Spend · My stack (Plan) · 89 complete rigs from 171 of 209 catalog parts · recomputed in 7404 ms

NVIDIA DGX Station GB300 (Grace Blackwell Ultra), 748GB is the most expensive machine this catalog can build, at $106,625 — first of 81 rigs ranked on landed cost. No rig in this catalog meets “My stack (Plan)”.

NO RIG MEETS THE PLANNO CONFIDENCE LEVEL

8,739 price observations on record, 4 of them carrying a shipping charge. 81 rigs ranked on total cost, 8 listed without a rank. Every number below expands to the arithmetic that produced it.

Since yesterdayNo change — the same rig leads this ranking and not one of the 81 rigs costable on both days moved.compared with prices observed on or before Oct 18, 11:01 PM ET (24 h ago)

What moved

81 of 89 rigs had a landed cost on both days; the rest were unpriced on one of them and are not counted as unchanged.
No rig's cost moved in that window.
The leader 24 h ago was NVIDIA DGX Station GB300 (Grace Blackwell Ultra), 748GB at $106,625.

How these numbers are costed

81 of 89 at landed cost · 81 of 89 at the price floor

Price + shipping + NJ sales tax, per SPEC §8.4. Unavailable for any offer whose listing does not state a shipping charge — unknown shipping is not free shipping. Landed cost is this screen's basis whenever a rig has one, and 81 of 89 do.

Cost basis

Landed cost · 81/89 Price floor · 81/89

Price + shipping + NJ sales tax, per SPEC §8.4. Unavailable for any offer whose listing does not state a shipping charge — unknown shipping is not free shipping. The screen chose this basis by the rule above the table. Landed cost is computable for 81 of 89 rigs; the price floor for 81 of 89.

Rank by

Total cost Cost per GB Cost per GB/s Cost per tok/s · GLM-5.3 Purchase + 5 yr power

Ranked by total cost, most expensive first — SPEC §15.11 target 1 ranks complete rigs by total landed cost descending.

Cost per tok/s measured on

GLM-5.3

The only decoding model in the active Plan. A rig that cannot load it has no cost per tok/s, and the cell says which band it fell in.

Complete rigs, ranked

81 ranked by total cost, largest first · 0 meet the Plan

Every complete rig the configurator can build from this catalog, ranked by total cost (largest first), with the cost basis, the three SPEC §8.4 value ratios and the provenance of each rig's price
# rank	Rig	Plan	Landed (the ranking key)	+ 5 yr power	$ / GB usable	$ / GB/s b/w	$ / tok/s decode
1	NVIDIA DGX Station GB300 (Grace Blackwell Ultra), 748GBsealed appliance · 1 part · break down its parts	4 BLOCKERS	$106,625 (derived) DERIVEDNVIDIA DGX Station GB300 (Grace Blackwell Ultra), 748GB: landed costArithmetic on observed inputs only. No tunable assumption enters it. rig_landed = Σ part landed_total $106,625 = $106,625 NVIDIA DGX Station GB300 (Grace Blackwell Ultra), 748GB 10662499 ¢ — Central Computers · ship Some part prices are older than 36 h. LOWER BOUND: 1 of 1 parts come from a retailer that does not state a delivery charge (Central Computers), so the rig costs at least this and never less.	$111,921 (estimate) ESTIMATENVIDIA DGX Station GB300 (Grace Blackwell Ultra), 748GB: purchase plus five years of powerArithmetic that also depends on a tunable assumption, listed below. purchase_plus_5yr_power = purchase + five_year_power purchase_plus_5yr_power = $106,624.99 + $5,295.68 = $111,920.67 purchase_cents 10662499 ¢ — rig_landed = Σ part landed_total five_year_power_cents 529568 ¢ — five_year_power = annual_kwh × ¢_per_kwh × 5 Some part prices are older than 36 h. LOWER BOUND: 1 of 1 parts come from a retailer that does not state a delivery charge (Central Computers), so the rig costs at least this and never less. Watts are at the wall (PSU conversion losses included by the caller). Load and idle draws are averages over their hours, not peaks. Duty cycle (Settings → Power): 6 h/day at load, 18 h/day idle. 365 days per year (leap days ignored, −0.07%). Flat rate for the whole period: no escalation, no time-of-use or tiered pricing, no demand charges. Rounded to the nearest cent, half-up, once from the unrounded annual cost.	$154.34 (derived) DERIVEDNVIDIA DGX Station GB300 (Grace Blackwell Ultra), 748GB: cost per GBArithmetic on observed inputs only. No tunable assumption enters it. cost_per_gb = landed_cost / usable_memory_gb = $106,625 / 690.84 GB = $154.34 /GB landed_cost 106624.99 USD — rig_landed = Σ part landed_total usable_memory_gb 690.84 GB — usable memory after the OS and framework reservation (Settings → Estimator) Some part prices are older than 36 h. LOWER BOUND: 1 of 1 parts come from a retailer that does not state a delivery charge (Central Computers), so the rig costs at least this and never less. Usable memory is below physical memory for at least one pool (OS/driver reservation; editable in Settings).	$15.02 (derived) DERIVEDNVIDIA DGX Station GB300 (Grace Blackwell Ultra), 748GB: cost per GB/sArithmetic on observed inputs only. No tunable assumption enters it. cost_per_gbs = landed_cost / memory_bandwidth_gbs = $106,625 / 7,100 GB/s = $15.02 per GB/s landed_cost 106624.99 USD — rig_landed = Σ part landed_total memory_bandwidth_gbs 7100 GB/s — the published bandwidth of the rig's fastest memory pool Some part prices are older than 36 h. LOWER BOUND: 1 of 1 parts come from a retailer that does not state a delivery charge (Central Computers), so the rig costs at least this and never less. Weights are read from the fastest pool first; a rig whose weights spill into a slower pool decodes below this figure.	$7,556 (estimate) ESTIMATENVIDIA DGX Station GB300 (Grace Blackwell Ultra), 748GB: cost per tok/sArithmetic that also depends on a tunable assumption, listed below. cost_per_tok_s = landed_cost / decode_tok_s = $106,625 / 14.11 tok/s = $7,555.89 per tok/s landed_cost 106624.99 USD — rig_landed = Σ part landed_total decode_tok_s 14.11 tok/s — GLM-5.3 decoding on this rig at the Plan's context Some part prices are older than 36 h. LOWER BOUND: 1 of 1 parts come from a retailer that does not state a delivery charge (Central Computers), so the rig costs at least this and never less. MoE: only active parameters are read per token (SPEC §3). Effective bits include embeddings and higher-precision tensors in mixed quants. Assumes active weights are quantized like the file average. KV cache is re-read on every token (DECISIONS D-008). Pools are read sequentially per token, so per-pool times add (harmonic combination). Per-token overhead on a split placement is DERIVED, not measured: each pool's own single-pool figure is charged in proportion to the share of each token it serves, because the fixed cost is per operation and the split puts that share of the operations on each device. Calibrated with that pool serving the whole model (D-013): GPU HBM3e (1.58 ms), CPU LPDDR5X (0 ms). Batch size 1 (single user, interactive).
Priced fromCentral Computerscondition newaccess tier 1, route jsonld_productstale 44 h — past the 36 hour freshness windowshipped
2	Hybrid · RTX PRO 6000 Blackwell Workstation Edition, 96GB + EPYC 9015 · 8× Server Premier KSM64R52BD4-128MB 128GB DDR5-6400 RDIMMDIY workstation · 7 parts · break down its parts	4 BLOCKERS	$71,596 (derived) DERIVEDHybrid · RTX PRO 6000 Blackwell Workstation Edition, 96GB + EPYC 9015 · 8× Server Premier KSM64R52BD4-128MB 128GB DDR5-6400 RDIMM: landed costArithmetic on observed inputs only. No tunable assumption enters it. rig_landed = Σ part landed_total $638 + $895 + $49,900 + $19,192 + $75 + $192 + $704 = $71,596 AMD EPYC 9015 63809 ¢ — SabrePC · ship ASRock Rack GENOAD8QM3-2T/BCM 89458 ¢ — Newegg · ship Kingston Server Premier KSM64R52BD4-128MB 128GB DDR5-6400 RDIMM 4990040 ¢ — Newegg · ship NVIDIA RTX PRO 6000 Blackwell Workstation Edition, 96GB 1919249 ¢ — Newegg · ship ARCTIC Freezer 4U-SP5 7463 ¢ — ARCTIC · ship Phanteks Enthoo Pro 2 Server Edition 19191 ¢ — Newegg · ship Seasonic PRIME PX-2200 70371 ¢ — Newegg · ship Some part prices are older than 36 h. LOWER BOUND: 7 of 7 parts come from a retailer that does not state a delivery charge (SabrePC, Newegg, Newegg, Newegg, ARCTIC, Newegg, Newegg), so the rig costs at least this and never less.	Unavailable — Purchase + 5-year power needs the power term. Load draw at the wall is unavailable: Electrical feasibility cannot be computed exactly: no load or TDP watts for ASRock Rack GENOAD8QM3-2T/BCM, Kingston Server Premier KSM64R52BD4-128MB 128GB DDR5-6400 RDIMM, ARCTIC Freezer 4U-SP5, Phanteks Enthoo Pro 2 Server Edition. (missing: ASRock Rack GENOAD8QM3-2T/BCM: load_watts or tdp_watts, Kingston Server Premier KSM64R52BD4-128MB 128GB DDR5-6400 RDIMM: load_watts or tdp_watts, ARCTIC Freezer 4U-SP5: load_watts or tdp_watts, Phanteks Enthoo Pro 2 Server Edition: load_watts or tdp_watts, AMD EPYC 9015: idle_watts, ASRock Rack GENOAD8QM3-2T/BCM: idle_watts, Kingston Server Premier KSM64R52BD4-128MB 128GB DDR5-6400 RDIMM: idle_watts, NVIDIA RTX PRO 6000 Blackwell Workstation Edition, 96GB: idle_watts, ARCTIC Freezer 4U-SP5: idle_watts, Phanteks Enthoo Pro 2 Server Edition: idle_watts, Seasonic PRIME PX-2200: idle_watts)	$67.17 (derived) DERIVEDHybrid · RTX PRO 6000 Blackwell Workstation Edition, 96GB + EPYC 9015 · 8× Server Premier KSM64R52BD4-128MB 128GB DDR5-6400 RDIMM: cost per GBArithmetic on observed inputs only. No tunable assumption enters it. cost_per_gb = landed_cost / usable_memory_gb = $71,596 / 1,065.92 GB = $67.17 /GB landed_cost 71595.81 USD — rig_landed = Σ part landed_total usable_memory_gb 1065.92 GB — usable memory after the OS and framework reservation (Settings → Estimator) Some part prices are older than 36 h. LOWER BOUND: 7 of 7 parts come from a retailer that does not state a delivery charge (SabrePC, Newegg, Newegg, Newegg, ARCTIC, Newegg, Newegg), so the rig costs at least this and never less. Usable memory is below physical memory for at least one pool (OS/driver reservation; editable in Settings).	$39.95 (derived) DERIVEDHybrid · RTX PRO 6000 Blackwell Workstation Edition, 96GB + EPYC 9015 · 8× Server Premier KSM64R52BD4-128MB 128GB DDR5-6400 RDIMM: cost per GB/sArithmetic on observed inputs only. No tunable assumption enters it. cost_per_gbs = landed_cost / memory_bandwidth_gbs = $71,596 / 1,792 GB/s = $39.95 per GB/s landed_cost 71595.81 USD — rig_landed = Σ part landed_total memory_bandwidth_gbs 1792 GB/s — the published bandwidth of the rig's fastest memory pool Some part prices are older than 36 h. LOWER BOUND: 7 of 7 parts come from a retailer that does not state a delivery charge (SabrePC, Newegg, Newegg, Newegg, ARCTIC, Newegg, Newegg), so the rig costs at least this and never less. Weights are read from the fastest pool first; a rig whose weights spill into a slower pool decodes below this figure.	$10,748 (estimate) ESTIMATEHybrid · RTX PRO 6000 Blackwell Workstation Edition, 96GB + EPYC 9015 · 8× Server Premier KSM64R52BD4-128MB 128GB DDR5-6400 RDIMM: cost per tok/sArithmetic that also depends on a tunable assumption, listed below. cost_per_tok_s = landed_cost / decode_tok_s = $71,596 / 6.66 tok/s = $10,747.87 per tok/s landed_cost 71595.81 USD — rig_landed = Σ part landed_total decode_tok_s 6.66 tok/s — GLM-5.3 decoding on this rig at the Plan's context Some part prices are older than 36 h. LOWER BOUND: 7 of 7 parts come from a retailer that does not state a delivery charge (SabrePC, Newegg, Newegg, Newegg, ARCTIC, Newegg, Newegg), so the rig costs at least this and never less. MoE: only active parameters are read per token (SPEC §3). Effective bits include embeddings and higher-precision tensors in mixed quants. Assumes active weights are quantized like the file average. KV cache is re-read on every token (DECISIONS D-008). Pools are read sequentially per token, so per-pool times add (harmonic combination). Per-token overhead on a split placement is DERIVED, not measured: each pool's own single-pool figure is charged in proportion to the share of each token it serves, because the fixed cost is per operation and the split puts that share of the operations on each device. Calibrated with that pool serving the whole model (D-013): GPU VRAM (1.58 ms), System RAM (44.61 ms). Batch size 1 (single user, interactive).
Priced from7 offers across 3 retailersoldeststale 44 h — past the 36 hour freshness windowopen the rig below for each part’s retailer, condition and route
3	Hybrid · RTX PRO 6000 Blackwell Max-Q Workstation Edition, 96GB + EPYC 9015 · 8× Server Premier KSM64R52BD4-128MB 128GB DDR5-6400 RDIMMDIY workstation · 7 parts · break down its parts	4 BLOCKERS	$65,124 (derived) DERIVEDHybrid · RTX PRO 6000 Blackwell Max-Q Workstation Edition, 96GB + EPYC 9015 · 8× Server Premier KSM64R52BD4-128MB 128GB DDR5-6400 RDIMM: landed costArithmetic on observed inputs only. No tunable assumption enters it. rig_landed = Σ part landed_total $638 + $895 + $49,900 + $12,721 + $75 + $192 + $704 = $65,124 AMD EPYC 9015 63809 ¢ — SabrePC · ship ASRock Rack GENOAD8QM3-2T/BCM 89458 ¢ — Newegg · ship Kingston Server Premier KSM64R52BD4-128MB 128GB DDR5-6400 RDIMM 4990040 ¢ — Newegg · ship NVIDIA RTX PRO 6000 Blackwell Max-Q Workstation Edition, 96GB 1272101 ¢ — SabrePC · ship ARCTIC Freezer 4U-SP5 7463 ¢ — ARCTIC · ship Phanteks Enthoo Pro 2 Server Edition 19191 ¢ — Newegg · ship Seasonic PRIME PX-2200 70371 ¢ — Newegg · ship Some part prices are older than 36 h. LOWER BOUND: 7 of 7 parts come from a retailer that does not state a delivery charge (SabrePC, Newegg, Newegg, SabrePC, ARCTIC, Newegg, Newegg), so the rig costs at least this and never less.	Unavailable — Purchase + 5-year power needs the power term. Load draw at the wall is unavailable: Electrical feasibility cannot be computed exactly: no load or TDP watts for ASRock Rack GENOAD8QM3-2T/BCM, Kingston Server Premier KSM64R52BD4-128MB 128GB DDR5-6400 RDIMM, ARCTIC Freezer 4U-SP5, Phanteks Enthoo Pro 2 Server Edition. (missing: ASRock Rack GENOAD8QM3-2T/BCM: load_watts or tdp_watts, Kingston Server Premier KSM64R52BD4-128MB 128GB DDR5-6400 RDIMM: load_watts or tdp_watts, ARCTIC Freezer 4U-SP5: load_watts or tdp_watts, Phanteks Enthoo Pro 2 Server Edition: load_watts or tdp_watts, AMD EPYC 9015: idle_watts, ASRock Rack GENOAD8QM3-2T/BCM: idle_watts, Kingston Server Premier KSM64R52BD4-128MB 128GB DDR5-6400 RDIMM: idle_watts, NVIDIA RTX PRO 6000 Blackwell Max-Q Workstation Edition, 96GB: idle_watts, ARCTIC Freezer 4U-SP5: idle_watts, Phanteks Enthoo Pro 2 Server Edition: idle_watts, Seasonic PRIME PX-2200: idle_watts)	$61.10 (derived) DERIVEDHybrid · RTX PRO 6000 Blackwell Max-Q Workstation Edition, 96GB + EPYC 9015 · 8× Server Premier KSM64R52BD4-128MB 128GB DDR5-6400 RDIMM: cost per GBArithmetic on observed inputs only. No tunable assumption enters it. cost_per_gb = landed_cost / usable_memory_gb = $65,124 / 1,065.92 GB = $61.1 /GB landed_cost 65124.33 USD — rig_landed = Σ part landed_total usable_memory_gb 1065.92 GB — usable memory after the OS and framework reservation (Settings → Estimator) Some part prices are older than 36 h. LOWER BOUND: 7 of 7 parts come from a retailer that does not state a delivery charge (SabrePC, Newegg, Newegg, SabrePC, ARCTIC, Newegg, Newegg), so the rig costs at least this and never less. Usable memory is below physical memory for at least one pool (OS/driver reservation; editable in Settings).	$36.34 (derived) DERIVEDHybrid · RTX PRO 6000 Blackwell Max-Q Workstation Edition, 96GB + EPYC 9015 · 8× Server Premier KSM64R52BD4-128MB 128GB DDR5-6400 RDIMM: cost per GB/sArithmetic on observed inputs only. No tunable assumption enters it. cost_per_gbs = landed_cost / memory_bandwidth_gbs = $65,124 / 1,792 GB/s = $36.34 per GB/s landed_cost 65124.33 USD — rig_landed = Σ part landed_total memory_bandwidth_gbs 1792 GB/s — the published bandwidth of the rig's fastest memory pool Some part prices are older than 36 h. LOWER BOUND: 7 of 7 parts come from a retailer that does not state a delivery charge (SabrePC, Newegg, Newegg, SabrePC, ARCTIC, Newegg, Newegg), so the rig costs at least this and never less. Weights are read from the fastest pool first; a rig whose weights spill into a slower pool decodes below this figure.	$9,776 (estimate) ESTIMATEHybrid · RTX PRO 6000 Blackwell Max-Q Workstation Edition, 96GB + EPYC 9015 · 8× Server Premier KSM64R52BD4-128MB 128GB DDR5-6400 RDIMM: cost per tok/sArithmetic that also depends on a tunable assumption, listed below. cost_per_tok_s = landed_cost / decode_tok_s = $65,124 / 6.66 tok/s = $9,776.38 per tok/s landed_cost 65124.33 USD — rig_landed = Σ part landed_total decode_tok_s 6.66 tok/s — GLM-5.3 decoding on this rig at the Plan's context Some part prices are older than 36 h. LOWER BOUND: 7 of 7 parts come from a retailer that does not state a delivery charge (SabrePC, Newegg, Newegg, SabrePC, ARCTIC, Newegg, Newegg), so the rig costs at least this and never less. MoE: only active parameters are read per token (SPEC §3). Effective bits include embeddings and higher-precision tensors in mixed quants. Assumes active weights are quantized like the file average. KV cache is re-read on every token (DECISIONS D-008). Pools are read sequentially per token, so per-pool times add (harmonic combination). Per-token overhead on a split placement is DERIVED, not measured: each pool's own single-pool figure is charged in proportion to the share of each token it serves, because the fixed cost is per operation and the split puts that share of the operations on each device. Calibrated with that pool serving the whole model (D-013): GPU VRAM (1.58 ms), System RAM (44.61 ms). Batch size 1 (single user, interactive).
Priced from7 offers across 3 retailersoldeststale 44 h — past the 36 hour freshness windowopen the rig below for each part’s retailer, condition and route
4	Hybrid · GeForce RTX 5090 32G GAMING TRIO OC + EPYC 9015 · 8× Server Premier KSM64R52BD4-128MB 128GB DDR5-6400 RDIMMDIY workstation · 7 parts · break down its parts	4 BLOCKERS	$57,841 (derived) DERIVEDHybrid · GeForce RTX 5090 32G GAMING TRIO OC + EPYC 9015 · 8× Server Premier KSM64R52BD4-128MB 128GB DDR5-6400 RDIMM: landed costArithmetic on observed inputs only. No tunable assumption enters it. rig_landed = Σ part landed_total $638 + $895 + $49,900 + $5,438 + $75 + $192 + $704 = $57,841 AMD EPYC 9015 63809 ¢ — SabrePC · ship ASRock Rack GENOAD8QM3-2T/BCM 89458 ¢ — Newegg · ship Kingston Server Premier KSM64R52BD4-128MB 128GB DDR5-6400 RDIMM 4990040 ¢ — Newegg · ship MSI GeForce RTX 5090 32G GAMING TRIO OC 543786 ¢ — Newegg · ship ARCTIC Freezer 4U-SP5 7463 ¢ — ARCTIC · ship Phanteks Enthoo Pro 2 Server Edition 19191 ¢ — Newegg · ship Seasonic PRIME PX-2200 70371 ¢ — Newegg · ship Some part prices are older than 36 h. LOWER BOUND: 7 of 7 parts come from a retailer that does not state a delivery charge (SabrePC, Newegg, Newegg, Newegg, ARCTIC, Newegg, Newegg), so the rig costs at least this and never less.	Unavailable — Purchase + 5-year power needs the power term. Load draw at the wall is unavailable: Electrical feasibility cannot be computed exactly: no load or TDP watts for ASRock Rack GENOAD8QM3-2T/BCM, Kingston Server Premier KSM64R52BD4-128MB 128GB DDR5-6400 RDIMM, ARCTIC Freezer 4U-SP5, Phanteks Enthoo Pro 2 Server Edition. (missing: ASRock Rack GENOAD8QM3-2T/BCM: load_watts or tdp_watts, Kingston Server Premier KSM64R52BD4-128MB 128GB DDR5-6400 RDIMM: load_watts or tdp_watts, ARCTIC Freezer 4U-SP5: load_watts or tdp_watts, Phanteks Enthoo Pro 2 Server Edition: load_watts or tdp_watts, AMD EPYC 9015: idle_watts, ASRock Rack GENOAD8QM3-2T/BCM: idle_watts, Kingston Server Premier KSM64R52BD4-128MB 128GB DDR5-6400 RDIMM: idle_watts, MSI GeForce RTX 5090 32G GAMING TRIO OC: idle_watts, ARCTIC Freezer 4U-SP5: idle_watts, Phanteks Enthoo Pro 2 Server Edition: idle_watts, Seasonic PRIME PX-2200: idle_watts)	$57.62 (derived) DERIVEDHybrid · GeForce RTX 5090 32G GAMING TRIO OC + EPYC 9015 · 8× Server Premier KSM64R52BD4-128MB 128GB DDR5-6400 RDIMM: cost per GBArithmetic on observed inputs only. No tunable assumption enters it. cost_per_gb = landed_cost / usable_memory_gb = $57,841 / 1,003.84 GB = $57.62 /GB landed_cost 57841.18 USD — rig_landed = Σ part landed_total usable_memory_gb 1003.84 GB — usable memory after the OS and framework reservation (Settings → Estimator) Some part prices are older than 36 h. LOWER BOUND: 7 of 7 parts come from a retailer that does not state a delivery charge (SabrePC, Newegg, Newegg, Newegg, ARCTIC, Newegg, Newegg), so the rig costs at least this and never less. Usable memory is below physical memory for at least one pool (OS/driver reservation; editable in Settings).	$32.28 (derived) DERIVEDHybrid · GeForce RTX 5090 32G GAMING TRIO OC + EPYC 9015 · 8× Server Premier KSM64R52BD4-128MB 128GB DDR5-6400 RDIMM: cost per GB/sArithmetic on observed inputs only. No tunable assumption enters it. cost_per_gbs = landed_cost / memory_bandwidth_gbs = $57,841 / 1,792 GB/s = $32.28 per GB/s landed_cost 57841.18 USD — rig_landed = Σ part landed_total memory_bandwidth_gbs 1792 GB/s — the published bandwidth of the rig's fastest memory pool Some part prices are older than 36 h. LOWER BOUND: 7 of 7 parts come from a retailer that does not state a delivery charge (SabrePC, Newegg, Newegg, Newegg, ARCTIC, Newegg, Newegg), so the rig costs at least this and never less. Weights are read from the fastest pool first; a rig whose weights spill into a slower pool decodes below this figure.	$9,685 (estimate) ESTIMATEHybrid · GeForce RTX 5090 32G GAMING TRIO OC + EPYC 9015 · 8× Server Premier KSM64R52BD4-128MB 128GB DDR5-6400 RDIMM: cost per tok/sArithmetic that also depends on a tunable assumption, listed below. cost_per_tok_s = landed_cost / decode_tok_s = $57,841 / 5.97 tok/s = $9,685.05 per tok/s landed_cost 57841.18 USD — rig_landed = Σ part landed_total decode_tok_s 5.97 tok/s — GLM-5.3 decoding on this rig at the Plan's context Some part prices are older than 36 h. LOWER BOUND: 7 of 7 parts come from a retailer that does not state a delivery charge (SabrePC, Newegg, Newegg, Newegg, ARCTIC, Newegg, Newegg), so the rig costs at least this and never less. MoE: only active parameters are read per token (SPEC §3). Effective bits include embeddings and higher-precision tensors in mixed quants. Assumes active weights are quantized like the file average. KV cache is re-read on every token (DECISIONS D-008). Pools are read sequentially per token, so per-pool times add (harmonic combination). Per-token overhead on a split placement is DERIVED, not measured: each pool's own single-pool figure is charged in proportion to the share of each token it serves, because the fixed cost is per operation and the split puts that share of the operations on each device. Calibrated with that pool serving the whole model (D-013): GPU VRAM (1.58 ms), System RAM (44.61 ms). Batch size 1 (single user, interactive).
Priced from7 offers across 3 retailersoldeststale 44 h — past the 36 hour freshness windowopen the rig below for each part’s retailer, condition and route
5	Hybrid · Radeon AI PRO R9700 32GB (32358-01-20G) + EPYC 9015 · 8× Server Premier KSM64R52BD4-128MB 128GB DDR5-6400 RDIMMDIY workstation · 7 parts · break down its parts	5 BLOCKERS	$54,323 (derived) DERIVEDHybrid · Radeon AI PRO R9700 32GB (32358-01-20G) + EPYC 9015 · 8× Server Premier KSM64R52BD4-128MB 128GB DDR5-6400 RDIMM: landed costArithmetic on observed inputs only. No tunable assumption enters it. rig_landed = Σ part landed_total $638 + $895 + $49,900 + $1,919 + $75 + $192 + $704 = $54,323 AMD EPYC 9015 63809 ¢ — SabrePC · ship ASRock Rack GENOAD8QM3-2T/BCM 89458 ¢ — Newegg · ship Kingston Server Premier KSM64R52BD4-128MB 128GB DDR5-6400 RDIMM 4990040 ¢ — Newegg · ship Sapphire Radeon AI PRO R9700 32GB (32358-01-20G) 191924 ¢ — Newegg · ship ARCTIC Freezer 4U-SP5 7463 ¢ — ARCTIC · ship Phanteks Enthoo Pro 2 Server Edition 19191 ¢ — Newegg · ship Seasonic PRIME PX-2200 70371 ¢ — Newegg · ship Some part prices are older than 36 h. LOWER BOUND: 7 of 7 parts come from a retailer that does not state a delivery charge (SabrePC, Newegg, Newegg, Newegg, ARCTIC, Newegg, Newegg), so the rig costs at least this and never less.	Unavailable — Purchase + 5-year power needs the power term. Load draw at the wall is unavailable: Electrical feasibility cannot be computed exactly: no load or TDP watts for ASRock Rack GENOAD8QM3-2T/BCM, Kingston Server Premier KSM64R52BD4-128MB 128GB DDR5-6400 RDIMM, ARCTIC Freezer 4U-SP5, Phanteks Enthoo Pro 2 Server Edition. (missing: ASRock Rack GENOAD8QM3-2T/BCM: load_watts or tdp_watts, Kingston Server Premier KSM64R52BD4-128MB 128GB DDR5-6400 RDIMM: load_watts or tdp_watts, ARCTIC Freezer 4U-SP5: load_watts or tdp_watts, Phanteks Enthoo Pro 2 Server Edition: load_watts or tdp_watts, AMD EPYC 9015: idle_watts, ASRock Rack GENOAD8QM3-2T/BCM: idle_watts, Kingston Server Premier KSM64R52BD4-128MB 128GB DDR5-6400 RDIMM: idle_watts, Sapphire Radeon AI PRO R9700 32GB (32358-01-20G): idle_watts, ARCTIC Freezer 4U-SP5: idle_watts, Phanteks Enthoo Pro 2 Server Edition: idle_watts, Seasonic PRIME PX-2200: idle_watts)	$54.11 (derived) DERIVEDHybrid · Radeon AI PRO R9700 32GB (32358-01-20G) + EPYC 9015 · 8× Server Premier KSM64R52BD4-128MB 128GB DDR5-6400 RDIMM: cost per GBArithmetic on observed inputs only. No tunable assumption enters it. cost_per_gb = landed_cost / usable_memory_gb = $54,323 / 1,003.84 GB = $54.11 /GB landed_cost 54322.56 USD — rig_landed = Σ part landed_total usable_memory_gb 1003.84 GB — usable memory after the OS and framework reservation (Settings → Estimator) Some part prices are older than 36 h. LOWER BOUND: 7 of 7 parts come from a retailer that does not state a delivery charge (SabrePC, Newegg, Newegg, Newegg, ARCTIC, Newegg, Newegg), so the rig costs at least this and never less. Usable memory is below physical memory for at least one pool (OS/driver reservation; editable in Settings).	$84.88 (derived) DERIVEDHybrid · Radeon AI PRO R9700 32GB (32358-01-20G) + EPYC 9015 · 8× Server Premier KSM64R52BD4-128MB 128GB DDR5-6400 RDIMM: cost per GB/sArithmetic on observed inputs only. No tunable assumption enters it. cost_per_gbs = landed_cost / memory_bandwidth_gbs = $54,323 / 640 GB/s = $84.88 per GB/s landed_cost 54322.56 USD — rig_landed = Σ part landed_total memory_bandwidth_gbs 640 GB/s — the published bandwidth of the rig's fastest memory pool Some part prices are older than 36 h. LOWER BOUND: 7 of 7 parts come from a retailer that does not state a delivery charge (SabrePC, Newegg, Newegg, Newegg, ARCTIC, Newegg, Newegg), so the rig costs at least this and never less. Weights are read from the fastest pool first; a rig whose weights spill into a slower pool decodes below this figure.	$9,292 (estimate) ESTIMATEHybrid · Radeon AI PRO R9700 32GB (32358-01-20G) + EPYC 9015 · 8× Server Premier KSM64R52BD4-128MB 128GB DDR5-6400 RDIMM: cost per tok/sArithmetic that also depends on a tunable assumption, listed below. cost_per_tok_s = landed_cost / decode_tok_s = $54,323 / 5.85 tok/s = $9,292.44 per tok/s landed_cost 54322.56 USD — rig_landed = Σ part landed_total decode_tok_s 5.85 tok/s — GLM-5.3 decoding on this rig at the Plan's context Some part prices are older than 36 h. LOWER BOUND: 7 of 7 parts come from a retailer that does not state a delivery charge (SabrePC, Newegg, Newegg, Newegg, ARCTIC, Newegg, Newegg), so the rig costs at least this and never less. MoE: only active parameters are read per token (SPEC §3). Effective bits include embeddings and higher-precision tensors in mixed quants. Assumes active weights are quantized like the file average. KV cache is re-read on every token (DECISIONS D-008). Pools are read sequentially per token, so per-pool times add (harmonic combination). Per-token overhead on a split placement is DERIVED, not measured: each pool's own single-pool figure is charged in proportion to the share of each token it serves, because the fixed cost is per operation and the split puts that share of the operations on each device. Calibrated with that pool serving the whole model (D-013): GPU VRAM (1.58 ms), System RAM (44.61 ms). Batch size 1 (single user, interactive).
Priced from7 offers across 3 retailersoldeststale 44 h — past the 36 hour freshness windowopen the rig below for each part’s retailer, condition and route
6	Hybrid · AI TOP Radeon AI PRO R9700 32GB (GV-R9700AI TOP-32GD) + EPYC 9015 · 8× Server Premier KSM64R52BD4-128MB 128GB DDR5-6400 RDIMMDIY workstation · 7 parts · break down its parts	5 BLOCKERS	$54,228 (derived) DERIVEDHybrid · AI TOP Radeon AI PRO R9700 32GB (GV-R9700AI TOP-32GD) + EPYC 9015 · 8× Server Premier KSM64R52BD4-128MB 128GB DDR5-6400 RDIMM: landed costArithmetic on observed inputs only. No tunable assumption enters it. rig_landed = Σ part landed_total $638 + $895 + $49,900 + $1,824 + $75 + $192 + $704 = $54,228 AMD EPYC 9015 63809 ¢ — SabrePC · ship ASRock Rack GENOAD8QM3-2T/BCM 89458 ¢ — Newegg · ship Kingston Server Premier KSM64R52BD4-128MB 128GB DDR5-6400 RDIMM 4990040 ¢ — Newegg · ship GIGABYTE AI TOP Radeon AI PRO R9700 32GB (GV-R9700AI TOP-32GD) 182424 ¢ — SabrePC · ship ARCTIC Freezer 4U-SP5 7463 ¢ — ARCTIC · ship Phanteks Enthoo Pro 2 Server Edition 19191 ¢ — Newegg · ship Seasonic PRIME PX-2200 70371 ¢ — Newegg · ship Some part prices are older than 36 h. LOWER BOUND: 7 of 7 parts come from a retailer that does not state a delivery charge (SabrePC, Newegg, Newegg, SabrePC, ARCTIC, Newegg, Newegg), so the rig costs at least this and never less.	Unavailable — Purchase + 5-year power needs the power term. Load draw at the wall is unavailable: Electrical feasibility cannot be computed exactly: no load or TDP watts for ASRock Rack GENOAD8QM3-2T/BCM, Kingston Server Premier KSM64R52BD4-128MB 128GB DDR5-6400 RDIMM, ARCTIC Freezer 4U-SP5, Phanteks Enthoo Pro 2 Server Edition. (missing: ASRock Rack GENOAD8QM3-2T/BCM: load_watts or tdp_watts, Kingston Server Premier KSM64R52BD4-128MB 128GB DDR5-6400 RDIMM: load_watts or tdp_watts, ARCTIC Freezer 4U-SP5: load_watts or tdp_watts, Phanteks Enthoo Pro 2 Server Edition: load_watts or tdp_watts, AMD EPYC 9015: idle_watts, ASRock Rack GENOAD8QM3-2T/BCM: idle_watts, Kingston Server Premier KSM64R52BD4-128MB 128GB DDR5-6400 RDIMM: idle_watts, GIGABYTE AI TOP Radeon AI PRO R9700 32GB (GV-R9700AI TOP-32GD): idle_watts, ARCTIC Freezer 4U-SP5: idle_watts, Phanteks Enthoo Pro 2 Server Edition: idle_watts, Seasonic PRIME PX-2200: idle_watts)	$54.02 (derived) DERIVEDHybrid · AI TOP Radeon AI PRO R9700 32GB (GV-R9700AI TOP-32GD) + EPYC 9015 · 8× Server Premier KSM64R52BD4-128MB 128GB DDR5-6400 RDIMM: cost per GBArithmetic on observed inputs only. No tunable assumption enters it. cost_per_gb = landed_cost / usable_memory_gb = $54,228 / 1,003.84 GB = $54.02 /GB landed_cost 54227.56 USD — rig_landed = Σ part landed_total usable_memory_gb 1003.84 GB — usable memory after the OS and framework reservation (Settings → Estimator) Some part prices are older than 36 h. LOWER BOUND: 7 of 7 parts come from a retailer that does not state a delivery charge (SabrePC, Newegg, Newegg, SabrePC, ARCTIC, Newegg, Newegg), so the rig costs at least this and never less. Usable memory is below physical memory for at least one pool (OS/driver reservation; editable in Settings).	$84.73 (derived) DERIVEDHybrid · AI TOP Radeon AI PRO R9700 32GB (GV-R9700AI TOP-32GD) + EPYC 9015 · 8× Server Premier KSM64R52BD4-128MB 128GB DDR5-6400 RDIMM: cost per GB/sArithmetic on observed inputs only. No tunable assumption enters it. cost_per_gbs = landed_cost / memory_bandwidth_gbs = $54,228 / 640 GB/s = $84.73 per GB/s landed_cost 54227.56 USD — rig_landed = Σ part landed_total memory_bandwidth_gbs 640 GB/s — the published bandwidth of the rig's fastest memory pool Some part prices are older than 36 h. LOWER BOUND: 7 of 7 parts come from a retailer that does not state a delivery charge (SabrePC, Newegg, Newegg, SabrePC, ARCTIC, Newegg, Newegg), so the rig costs at least this and never less. Weights are read from the fastest pool first; a rig whose weights spill into a slower pool decodes below this figure.	$9,276 (estimate) ESTIMATEHybrid · AI TOP Radeon AI PRO R9700 32GB (GV-R9700AI TOP-32GD) + EPYC 9015 · 8× Server Premier KSM64R52BD4-128MB 128GB DDR5-6400 RDIMM: cost per tok/sArithmetic that also depends on a tunable assumption, listed below. cost_per_tok_s = landed_cost / decode_tok_s = $54,228 / 5.85 tok/s = $9,276.19 per tok/s landed_cost 54227.56 USD — rig_landed = Σ part landed_total decode_tok_s 5.85 tok/s — GLM-5.3 decoding on this rig at the Plan's context Some part prices are older than 36 h. LOWER BOUND: 7 of 7 parts come from a retailer that does not state a delivery charge (SabrePC, Newegg, Newegg, SabrePC, ARCTIC, Newegg, Newegg), so the rig costs at least this and never less. MoE: only active parameters are read per token (SPEC §3). Effective bits include embeddings and higher-precision tensors in mixed quants. Assumes active weights are quantized like the file average. KV cache is re-read on every token (DECISIONS D-008). Pools are read sequentially per token, so per-pool times add (harmonic combination). Per-token overhead on a split placement is DERIVED, not measured: each pool's own single-pool figure is charged in proportion to the share of each token it serves, because the fixed cost is per operation and the split puts that share of the operations on each device. Calibrated with that pool serving the whole model (D-013): GPU VRAM (1.58 ms), System RAM (44.61 ms). Batch size 1 (single user, interactive).
Priced from7 offers across 3 retailersoldeststale 44 h — past the 36 hour freshness windowopen the rig below for each part’s retailer, condition and route
7	Hybrid · GeForce RTX 5080 WINDFORCE OC 16G (GV-N5080WF3OC-16GD) + EPYC 9015 · 8× Server Premier KSM64R52BD4-128MB 128GB DDR5-6400 RDIMMDIY workstation · 7 parts · break down its parts	4 BLOCKERS	$54,216 (derived) DERIVEDHybrid · GeForce RTX 5080 WINDFORCE OC 16G (GV-N5080WF3OC-16GD) + EPYC 9015 · 8× Server Premier KSM64R52BD4-128MB 128GB DDR5-6400 RDIMM: landed costArithmetic on observed inputs only. No tunable assumption enters it. rig_landed = Σ part landed_total $638 + $895 + $49,900 + $1,813 + $75 + $192 + $704 = $54,216 AMD EPYC 9015 63809 ¢ — SabrePC · ship ASRock Rack GENOAD8QM3-2T/BCM 89458 ¢ — Newegg · ship Kingston Server Premier KSM64R52BD4-128MB 128GB DDR5-6400 RDIMM 4990040 ¢ — Newegg · ship GIGABYTE GeForce RTX 5080 WINDFORCE OC 16G (GV-N5080WF3OC-16GD) 181261 ¢ — Newegg · ship ARCTIC Freezer 4U-SP5 7463 ¢ — ARCTIC · ship Phanteks Enthoo Pro 2 Server Edition 19191 ¢ — Newegg · ship Seasonic PRIME PX-2200 70371 ¢ — Newegg · ship Some part prices are older than 36 h. LOWER BOUND: 7 of 7 parts come from a retailer that does not state a delivery charge (SabrePC, Newegg, Newegg, Newegg, ARCTIC, Newegg, Newegg), so the rig costs at least this and never less.	Unavailable — Purchase + 5-year power needs the power term. Load draw at the wall is unavailable: Electrical feasibility cannot be computed exactly: no load or TDP watts for ASRock Rack GENOAD8QM3-2T/BCM, Kingston Server Premier KSM64R52BD4-128MB 128GB DDR5-6400 RDIMM, ARCTIC Freezer 4U-SP5, Phanteks Enthoo Pro 2 Server Edition. (missing: ASRock Rack GENOAD8QM3-2T/BCM: load_watts or tdp_watts, Kingston Server Premier KSM64R52BD4-128MB 128GB DDR5-6400 RDIMM: load_watts or tdp_watts, ARCTIC Freezer 4U-SP5: load_watts or tdp_watts, Phanteks Enthoo Pro 2 Server Edition: load_watts or tdp_watts, AMD EPYC 9015: idle_watts, ASRock Rack GENOAD8QM3-2T/BCM: idle_watts, Kingston Server Premier KSM64R52BD4-128MB 128GB DDR5-6400 RDIMM: idle_watts, GIGABYTE GeForce RTX 5080 WINDFORCE OC 16G (GV-N5080WF3OC-16GD): idle_watts, ARCTIC Freezer 4U-SP5: idle_watts, Phanteks Enthoo Pro 2 Server Edition: idle_watts, Seasonic PRIME PX-2200: idle_watts)	$54.86 (derived) DERIVEDHybrid · GeForce RTX 5080 WINDFORCE OC 16G (GV-N5080WF3OC-16GD) + EPYC 9015 · 8× Server Premier KSM64R52BD4-128MB 128GB DDR5-6400 RDIMM: cost per GBArithmetic on observed inputs only. No tunable assumption enters it. cost_per_gb = landed_cost / usable_memory_gb = $54,216 / 988.32 GB = $54.86 /GB landed_cost 54215.93 USD — rig_landed = Σ part landed_total usable_memory_gb 988.32 GB — usable memory after the OS and framework reservation (Settings → Estimator) Some part prices are older than 36 h. LOWER BOUND: 7 of 7 parts come from a retailer that does not state a delivery charge (SabrePC, Newegg, Newegg, Newegg, ARCTIC, Newegg, Newegg), so the rig costs at least this and never less. Usable memory is below physical memory for at least one pool (OS/driver reservation; editable in Settings).	$56.47 (derived) DERIVEDHybrid · GeForce RTX 5080 WINDFORCE OC 16G (GV-N5080WF3OC-16GD) + EPYC 9015 · 8× Server Premier KSM64R52BD4-128MB 128GB DDR5-6400 RDIMM: cost per GB/sArithmetic on observed inputs only. No tunable assumption enters it. cost_per_gbs = landed_cost / memory_bandwidth_gbs = $54,216 / 960 GB/s = $56.47 per GB/s landed_cost 54215.93 USD — rig_landed = Σ part landed_total memory_bandwidth_gbs 960 GB/s — the published bandwidth of the rig's fastest memory pool Some part prices are older than 36 h. LOWER BOUND: 7 of 7 parts come from a retailer that does not state a delivery charge (SabrePC, Newegg, Newegg, Newegg, ARCTIC, Newegg, Newegg), so the rig costs at least this and never less. Weights are read from the fastest pool first; a rig whose weights spill into a slower pool decodes below this figure.	$9,360 (estimate) ESTIMATEHybrid · GeForce RTX 5080 WINDFORCE OC 16G (GV-N5080WF3OC-16GD) + EPYC 9015 · 8× Server Premier KSM64R52BD4-128MB 128GB DDR5-6400 RDIMM: cost per tok/sArithmetic that also depends on a tunable assumption, listed below. cost_per_tok_s = landed_cost / decode_tok_s = $54,216 / 5.79 tok/s = $9,360.06 per tok/s landed_cost 54215.93 USD — rig_landed = Σ part landed_total decode_tok_s 5.79 tok/s — GLM-5.3 decoding on this rig at the Plan's context Some part prices are older than 36 h. LOWER BOUND: 7 of 7 parts come from a retailer that does not state a delivery charge (SabrePC, Newegg, Newegg, Newegg, ARCTIC, Newegg, Newegg), so the rig costs at least this and never less. MoE: only active parameters are read per token (SPEC §3). Effective bits include embeddings and higher-precision tensors in mixed quants. Assumes active weights are quantized like the file average. KV cache is re-read on every token (DECISIONS D-008). Pools are read sequentially per token, so per-pool times add (harmonic combination). Per-token overhead on a split placement is DERIVED, not measured: each pool's own single-pool figure is charged in proportion to the share of each token it serves, because the fixed cost is per operation and the split puts that share of the operations on each device. Calibrated with that pool serving the whole model (D-013): GPU VRAM (1.58 ms), System RAM (44.61 ms). Batch size 1 (single user, interactive).
Priced from7 offers across 3 retailersoldeststale 44 h — past the 36 hour freshness windowopen the rig below for each part’s retailer, condition and route
8	Hybrid · Radeon AI PRO R9700 Creator 32GB (R9700 CT) + EPYC 9015 · 8× Server Premier KSM64R52BD4-128MB 128GB DDR5-6400 RDIMMDIY workstation · 7 parts · break down its parts	5 BLOCKERS	$54,216 (derived) DERIVEDHybrid · Radeon AI PRO R9700 Creator 32GB (R9700 CT) + EPYC 9015 · 8× Server Premier KSM64R52BD4-128MB 128GB DDR5-6400 RDIMM: landed costArithmetic on observed inputs only. No tunable assumption enters it. rig_landed = Σ part landed_total $638 + $895 + $49,900 + $1,813 + $75 + $192 + $704 = $54,216 AMD EPYC 9015 63809 ¢ — SabrePC · ship ASRock Rack GENOAD8QM3-2T/BCM 89458 ¢ — Newegg · ship Kingston Server Premier KSM64R52BD4-128MB 128GB DDR5-6400 RDIMM 4990040 ¢ — Newegg · ship ASRock Radeon AI PRO R9700 Creator 32GB (R9700 CT) 181261 ¢ — Newegg · ship ARCTIC Freezer 4U-SP5 7463 ¢ — ARCTIC · ship Phanteks Enthoo Pro 2 Server Edition 19191 ¢ — Newegg · ship Seasonic PRIME PX-2200 70371 ¢ — Newegg · ship Some part prices are older than 36 h. LOWER BOUND: 7 of 7 parts come from a retailer that does not state a delivery charge (SabrePC, Newegg, Newegg, Newegg, ARCTIC, Newegg, Newegg), so the rig costs at least this and never less.	Unavailable — Purchase + 5-year power needs the power term. Load draw at the wall is unavailable: Electrical feasibility cannot be computed exactly: no load or TDP watts for ASRock Rack GENOAD8QM3-2T/BCM, Kingston Server Premier KSM64R52BD4-128MB 128GB DDR5-6400 RDIMM, ARCTIC Freezer 4U-SP5, Phanteks Enthoo Pro 2 Server Edition. (missing: ASRock Rack GENOAD8QM3-2T/BCM: load_watts or tdp_watts, Kingston Server Premier KSM64R52BD4-128MB 128GB DDR5-6400 RDIMM: load_watts or tdp_watts, ARCTIC Freezer 4U-SP5: load_watts or tdp_watts, Phanteks Enthoo Pro 2 Server Edition: load_watts or tdp_watts, AMD EPYC 9015: idle_watts, ASRock Rack GENOAD8QM3-2T/BCM: idle_watts, Kingston Server Premier KSM64R52BD4-128MB 128GB DDR5-6400 RDIMM: idle_watts, ASRock Radeon AI PRO R9700 Creator 32GB (R9700 CT): idle_watts, ARCTIC Freezer 4U-SP5: idle_watts, Phanteks Enthoo Pro 2 Server Edition: idle_watts, Seasonic PRIME PX-2200: idle_watts)	$54.01 (derived) DERIVEDHybrid · Radeon AI PRO R9700 Creator 32GB (R9700 CT) + EPYC 9015 · 8× Server Premier KSM64R52BD4-128MB 128GB DDR5-6400 RDIMM: cost per GBArithmetic on observed inputs only. No tunable assumption enters it. cost_per_gb = landed_cost / usable_memory_gb = $54,216 / 1,003.84 GB = $54.01 /GB landed_cost 54215.93 USD — rig_landed = Σ part landed_total usable_memory_gb 1003.84 GB — usable memory after the OS and framework reservation (Settings → Estimator) Some part prices are older than 36 h. LOWER BOUND: 7 of 7 parts come from a retailer that does not state a delivery charge (SabrePC, Newegg, Newegg, Newegg, ARCTIC, Newegg, Newegg), so the rig costs at least this and never less. Usable memory is below physical memory for at least one pool (OS/driver reservation; editable in Settings).	$84.71 (derived) DERIVEDHybrid · Radeon AI PRO R9700 Creator 32GB (R9700 CT) + EPYC 9015 · 8× Server Premier KSM64R52BD4-128MB 128GB DDR5-6400 RDIMM: cost per GB/sArithmetic on observed inputs only. No tunable assumption enters it. cost_per_gbs = landed_cost / memory_bandwidth_gbs = $54,216 / 640 GB/s = $84.71 per GB/s landed_cost 54215.93 USD — rig_landed = Σ part landed_total memory_bandwidth_gbs 640 GB/s — the published bandwidth of the rig's fastest memory pool Some part prices are older than 36 h. LOWER BOUND: 7 of 7 parts come from a retailer that does not state a delivery charge (SabrePC, Newegg, Newegg, Newegg, ARCTIC, Newegg, Newegg), so the rig costs at least this and never less. Weights are read from the fastest pool first; a rig whose weights spill into a slower pool decodes below this figure.	$9,274 (estimate) ESTIMATEHybrid · Radeon AI PRO R9700 Creator 32GB (R9700 CT) + EPYC 9015 · 8× Server Premier KSM64R52BD4-128MB 128GB DDR5-6400 RDIMM: cost per tok/sArithmetic that also depends on a tunable assumption, listed below. cost_per_tok_s = landed_cost / decode_tok_s = $54,216 / 5.85 tok/s = $9,274.2 per tok/s landed_cost 54215.93 USD — rig_landed = Σ part landed_total decode_tok_s 5.85 tok/s — GLM-5.3 decoding on this rig at the Plan's context Some part prices are older than 36 h. LOWER BOUND: 7 of 7 parts come from a retailer that does not state a delivery charge (SabrePC, Newegg, Newegg, Newegg, ARCTIC, Newegg, Newegg), so the rig costs at least this and never less. MoE: only active parameters are read per token (SPEC §3). Effective bits include embeddings and higher-precision tensors in mixed quants. Assumes active weights are quantized like the file average. KV cache is re-read on every token (DECISIONS D-008). Pools are read sequentially per token, so per-pool times add (harmonic combination). Per-token overhead on a split placement is DERIVED, not measured: each pool's own single-pool figure is charged in proportion to the share of each token it serves, because the fixed cost is per operation and the split puts that share of the operations on each device. Calibrated with that pool serving the whole model (D-013): GPU VRAM (1.58 ms), System RAM (44.61 ms). Batch size 1 (single user, interactive).
Priced from7 offers across 3 retailersoldeststale 30 d — past the 36 hour freshness windowopen the rig below for each part’s retailer, condition and route
9	Hybrid · GeForce RTX 5080 16G SHADOW 3X OC + EPYC 9015 · 8× Server Premier KSM64R52BD4-128MB 128GB DDR5-6400 RDIMMDIY workstation · 7 parts · break down its parts	4 BLOCKERS	$54,205 (derived) DERIVEDHybrid · GeForce RTX 5080 16G SHADOW 3X OC + EPYC 9015 · 8× Server Premier KSM64R52BD4-128MB 128GB DDR5-6400 RDIMM: landed costArithmetic on observed inputs only. No tunable assumption enters it. rig_landed = Σ part landed_total $638 + $895 + $49,900 + $1,802 + $75 + $192 + $704 = $54,205 AMD EPYC 9015 63809 ¢ — SabrePC · ship ASRock Rack GENOAD8QM3-2T/BCM 89458 ¢ — Newegg · ship Kingston Server Premier KSM64R52BD4-128MB 128GB DDR5-6400 RDIMM 4990040 ¢ — Newegg · ship MSI GeForce RTX 5080 16G SHADOW 3X OC 180195 ¢ — Newegg · ship ARCTIC Freezer 4U-SP5 7463 ¢ — ARCTIC · ship Phanteks Enthoo Pro 2 Server Edition 19191 ¢ — Newegg · ship Seasonic PRIME PX-2200 70371 ¢ — Newegg · ship Some part prices are older than 36 h. LOWER BOUND: 7 of 7 parts come from a retailer that does not state a delivery charge (SabrePC, Newegg, Newegg, Newegg, ARCTIC, Newegg, Newegg), so the rig costs at least this and never less.	Unavailable — Purchase + 5-year power needs the power term. Load draw at the wall is unavailable: Electrical feasibility cannot be computed exactly: no load or TDP watts for ASRock Rack GENOAD8QM3-2T/BCM, Kingston Server Premier KSM64R52BD4-128MB 128GB DDR5-6400 RDIMM, ARCTIC Freezer 4U-SP5, Phanteks Enthoo Pro 2 Server Edition. (missing: ASRock Rack GENOAD8QM3-2T/BCM: load_watts or tdp_watts, Kingston Server Premier KSM64R52BD4-128MB 128GB DDR5-6400 RDIMM: load_watts or tdp_watts, ARCTIC Freezer 4U-SP5: load_watts or tdp_watts, Phanteks Enthoo Pro 2 Server Edition: load_watts or tdp_watts, AMD EPYC 9015: idle_watts, ASRock Rack GENOAD8QM3-2T/BCM: idle_watts, Kingston Server Premier KSM64R52BD4-128MB 128GB DDR5-6400 RDIMM: idle_watts, MSI GeForce RTX 5080 16G SHADOW 3X OC: idle_watts, ARCTIC Freezer 4U-SP5: idle_watts, Phanteks Enthoo Pro 2 Server Edition: idle_watts, Seasonic PRIME PX-2200: idle_watts)	$54.85 (derived) DERIVEDHybrid · GeForce RTX 5080 16G SHADOW 3X OC + EPYC 9015 · 8× Server Premier KSM64R52BD4-128MB 128GB DDR5-6400 RDIMM: cost per GBArithmetic on observed inputs only. No tunable assumption enters it. cost_per_gb = landed_cost / usable_memory_gb = $54,205 / 988.32 GB = $54.85 /GB landed_cost 54205.27 USD — rig_landed = Σ part landed_total usable_memory_gb 988.32 GB — usable memory after the OS and framework reservation (Settings → Estimator) Some part prices are older than 36 h. LOWER BOUND: 7 of 7 parts come from a retailer that does not state a delivery charge (SabrePC, Newegg, Newegg, Newegg, ARCTIC, Newegg, Newegg), so the rig costs at least this and never less. Usable memory is below physical memory for at least one pool (OS/driver reservation; editable in Settings).	$56.46 (derived) DERIVEDHybrid · GeForce RTX 5080 16G SHADOW 3X OC + EPYC 9015 · 8× Server Premier KSM64R52BD4-128MB 128GB DDR5-6400 RDIMM: cost per GB/sArithmetic on observed inputs only. No tunable assumption enters it. cost_per_gbs = landed_cost / memory_bandwidth_gbs = $54,205 / 960 GB/s = $56.46 per GB/s landed_cost 54205.27 USD — rig_landed = Σ part landed_total memory_bandwidth_gbs 960 GB/s — the published bandwidth of the rig's fastest memory pool Some part prices are older than 36 h. LOWER BOUND: 7 of 7 parts come from a retailer that does not state a delivery charge (SabrePC, Newegg, Newegg, Newegg, ARCTIC, Newegg, Newegg), so the rig costs at least this and never less. Weights are read from the fastest pool first; a rig whose weights spill into a slower pool decodes below this figure.	$9,358 (estimate) ESTIMATEHybrid · GeForce RTX 5080 16G SHADOW 3X OC + EPYC 9015 · 8× Server Premier KSM64R52BD4-128MB 128GB DDR5-6400 RDIMM: cost per tok/sArithmetic that also depends on a tunable assumption, listed below. cost_per_tok_s = landed_cost / decode_tok_s = $54,205 / 5.79 tok/s = $9,358.22 per tok/s landed_cost 54205.27 USD — rig_landed = Σ part landed_total decode_tok_s 5.79 tok/s — GLM-5.3 decoding on this rig at the Plan's context Some part prices are older than 36 h. LOWER BOUND: 7 of 7 parts come from a retailer that does not state a delivery charge (SabrePC, Newegg, Newegg, Newegg, ARCTIC, Newegg, Newegg), so the rig costs at least this and never less. MoE: only active parameters are read per token (SPEC §3). Effective bits include embeddings and higher-precision tensors in mixed quants. Assumes active weights are quantized like the file average. KV cache is re-read on every token (DECISIONS D-008). Pools are read sequentially per token, so per-pool times add (harmonic combination). Per-token overhead on a split placement is DERIVED, not measured: each pool's own single-pool figure is charged in proportion to the share of each token it serves, because the fixed cost is per operation and the split puts that share of the operations on each device. Calibrated with that pool serving the whole model (D-013): GPU VRAM (1.58 ms), System RAM (44.61 ms). Batch size 1 (single user, interactive).
Priced from7 offers across 3 retailersoldeststale 44 h — past the 36 hour freshness windowopen the rig below for each part’s retailer, condition and route
10	Hybrid · RTX PRO 6000 Blackwell Workstation Edition, 96GB + EPYC 9015 · 8× Server Premier KSM64R52BD4-96MC 96GB DDR5-6400 RDIMMDIY workstation · 7 parts · break down its parts	4 BLOCKERS	$54,109 (derived) DERIVEDHybrid · RTX PRO 6000 Blackwell Workstation Edition, 96GB + EPYC 9015 · 8× Server Premier KSM64R52BD4-96MC 96GB DDR5-6400 RDIMM: landed costArithmetic on observed inputs only. No tunable assumption enters it. rig_landed = Σ part landed_total $638 + $895 + $32,414 + $19,192 + $75 + $192 + $704 = $54,109 AMD EPYC 9015 63809 ¢ — SabrePC · ship ASRock Rack GENOAD8QM3-2T/BCM 89458 ¢ — Newegg · ship Kingston Server Premier KSM64R52BD4-96MC 96GB DDR5-6400 RDIMM 3241392 ¢ — Newegg · ship NVIDIA RTX PRO 6000 Blackwell Workstation Edition, 96GB 1919249 ¢ — Newegg · ship ARCTIC Freezer 4U-SP5 7463 ¢ — ARCTIC · ship Phanteks Enthoo Pro 2 Server Edition 19191 ¢ — Newegg · ship Seasonic PRIME PX-2200 70371 ¢ — Newegg · ship Some part prices are older than 36 h. LOWER BOUND: 7 of 7 parts come from a retailer that does not state a delivery charge (SabrePC, Newegg, Newegg, Newegg, ARCTIC, Newegg, Newegg), so the rig costs at least this and never less.	Unavailable — Purchase + 5-year power needs the power term. Load draw at the wall is unavailable: Electrical feasibility cannot be computed exactly: no load or TDP watts for ASRock Rack GENOAD8QM3-2T/BCM, Kingston Server Premier KSM64R52BD4-96MC 96GB DDR5-6400 RDIMM, ARCTIC Freezer 4U-SP5, Phanteks Enthoo Pro 2 Server Edition. (missing: ASRock Rack GENOAD8QM3-2T/BCM: load_watts or tdp_watts, Kingston Server Premier KSM64R52BD4-96MC 96GB DDR5-6400 RDIMM: load_watts or tdp_watts, ARCTIC Freezer 4U-SP5: load_watts or tdp_watts, Phanteks Enthoo Pro 2 Server Edition: load_watts or tdp_watts, AMD EPYC 9015: idle_watts, ASRock Rack GENOAD8QM3-2T/BCM: idle_watts, Kingston Server Premier KSM64R52BD4-96MC 96GB DDR5-6400 RDIMM: idle_watts, NVIDIA RTX PRO 6000 Blackwell Workstation Edition, 96GB: idle_watts, ARCTIC Freezer 4U-SP5: idle_watts, Phanteks Enthoo Pro 2 Server Edition: idle_watts, Seasonic PRIME PX-2200: idle_watts)	$65.77 (derived) DERIVEDHybrid · RTX PRO 6000 Blackwell Workstation Edition, 96GB + EPYC 9015 · 8× Server Premier KSM64R52BD4-96MC 96GB DDR5-6400 RDIMM: cost per GBArithmetic on observed inputs only. No tunable assumption enters it. cost_per_gb = landed_cost / usable_memory_gb = $54,109 / 822.72 GB = $65.77 /GB landed_cost 54109.33 USD — rig_landed = Σ part landed_total usable_memory_gb 822.72 GB — usable memory after the OS and framework reservation (Settings → Estimator) Some part prices are older than 36 h. LOWER BOUND: 7 of 7 parts come from a retailer that does not state a delivery charge (SabrePC, Newegg, Newegg, Newegg, ARCTIC, Newegg, Newegg), so the rig costs at least this and never less. Usable memory is below physical memory for at least one pool (OS/driver reservation; editable in Settings).	$30.19 (derived) DERIVEDHybrid · RTX PRO 6000 Blackwell Workstation Edition, 96GB + EPYC 9015 · 8× Server Premier KSM64R52BD4-96MC 96GB DDR5-6400 RDIMM: cost per GB/sArithmetic on observed inputs only. No tunable assumption enters it. cost_per_gbs = landed_cost / memory_bandwidth_gbs = $54,109 / 1,792 GB/s = $30.19 per GB/s landed_cost 54109.33 USD — rig_landed = Σ part landed_total memory_bandwidth_gbs 1792 GB/s — the published bandwidth of the rig's fastest memory pool Some part prices are older than 36 h. LOWER BOUND: 7 of 7 parts come from a retailer that does not state a delivery charge (SabrePC, Newegg, Newegg, Newegg, ARCTIC, Newegg, Newegg), so the rig costs at least this and never less. Weights are read from the fastest pool first; a rig whose weights spill into a slower pool decodes below this figure.	$8,123 (estimate) ESTIMATEHybrid · RTX PRO 6000 Blackwell Workstation Edition, 96GB + EPYC 9015 · 8× Server Premier KSM64R52BD4-96MC 96GB DDR5-6400 RDIMM: cost per tok/sArithmetic that also depends on a tunable assumption, listed below. cost_per_tok_s = landed_cost / decode_tok_s = $54,109 / 6.66 tok/s = $8,122.83 per tok/s landed_cost 54109.33 USD — rig_landed = Σ part landed_total decode_tok_s 6.66 tok/s — GLM-5.3 decoding on this rig at the Plan's context Some part prices are older than 36 h. LOWER BOUND: 7 of 7 parts come from a retailer that does not state a delivery charge (SabrePC, Newegg, Newegg, Newegg, ARCTIC, Newegg, Newegg), so the rig costs at least this and never less. MoE: only active parameters are read per token (SPEC §3). Effective bits include embeddings and higher-precision tensors in mixed quants. Assumes active weights are quantized like the file average. KV cache is re-read on every token (DECISIONS D-008). Pools are read sequentially per token, so per-pool times add (harmonic combination). Per-token overhead on a split placement is DERIVED, not measured: each pool's own single-pool figure is charged in proportion to the share of each token it serves, because the fixed cost is per operation and the split puts that share of the operations on each device. Calibrated with that pool serving the whole model (D-013): GPU VRAM (1.58 ms), System RAM (44.61 ms). Batch size 1 (single user, interactive).
Priced from7 offers across 3 retailersoldeststale 44 h — past the 36 hour freshness windowopen the rig below for each part’s retailer, condition and route
11	CPU · EPYC 9015 · 8× Server Premier KSM64R52BD4-128MB 128GB DDR5-6400 RDIMMDIY workstation · 6 parts · break down its parts	5 BLOCKERS	$52,403 (derived) DERIVEDCPU · EPYC 9015 · 8× Server Premier KSM64R52BD4-128MB 128GB DDR5-6400 RDIMM: landed costArithmetic on observed inputs only. No tunable assumption enters it. rig_landed = Σ part landed_total $638 + $895 + $49,900 + $75 + $192 + $704 = $52,403 AMD EPYC 9015 63809 ¢ — SabrePC · ship ASRock Rack GENOAD8QM3-2T/BCM 89458 ¢ — Newegg · ship Kingston Server Premier KSM64R52BD4-128MB 128GB DDR5-6400 RDIMM 4990040 ¢ — Newegg · ship ARCTIC Freezer 4U-SP5 7463 ¢ — ARCTIC · ship Phanteks Enthoo Pro 2 Server Edition 19191 ¢ — Newegg · ship Seasonic PRIME PX-2200 70371 ¢ — Newegg · ship Some part prices are older than 36 h. LOWER BOUND: 6 of 6 parts come from a retailer that does not state a delivery charge (SabrePC, Newegg, Newegg, ARCTIC, Newegg, Newegg), so the rig costs at least this and never less.	Unavailable — Purchase + 5-year power needs the power term. Load draw at the wall is unavailable: Electrical feasibility cannot be computed exactly: no load or TDP watts for ASRock Rack GENOAD8QM3-2T/BCM, Kingston Server Premier KSM64R52BD4-128MB 128GB DDR5-6400 RDIMM, ARCTIC Freezer 4U-SP5, Phanteks Enthoo Pro 2 Server Edition. (missing: ASRock Rack GENOAD8QM3-2T/BCM: load_watts or tdp_watts, Kingston Server Premier KSM64R52BD4-128MB 128GB DDR5-6400 RDIMM: load_watts or tdp_watts, ARCTIC Freezer 4U-SP5: load_watts or tdp_watts, Phanteks Enthoo Pro 2 Server Edition: load_watts or tdp_watts, AMD EPYC 9015: idle_watts, ASRock Rack GENOAD8QM3-2T/BCM: idle_watts, Kingston Server Premier KSM64R52BD4-128MB 128GB DDR5-6400 RDIMM: idle_watts, ARCTIC Freezer 4U-SP5: idle_watts, Phanteks Enthoo Pro 2 Server Edition: idle_watts, Seasonic PRIME PX-2200: idle_watts)	$53.87 (derived) DERIVEDCPU · EPYC 9015 · 8× Server Premier KSM64R52BD4-128MB 128GB DDR5-6400 RDIMM: cost per GBArithmetic on observed inputs only. No tunable assumption enters it. cost_per_gb = landed_cost / usable_memory_gb = $52,403 / 972.8 GB = $53.87 /GB landed_cost 52403.32 USD — rig_landed = Σ part landed_total usable_memory_gb 972.8 GB — usable memory after the OS and framework reservation (Settings → Estimator) Some part prices are older than 36 h. LOWER BOUND: 6 of 6 parts come from a retailer that does not state a delivery charge (SabrePC, Newegg, Newegg, ARCTIC, Newegg, Newegg), so the rig costs at least this and never less. Usable memory is below physical memory for at least one pool (OS/driver reservation; editable in Settings).	$170.58 (derived) DERIVEDCPU · EPYC 9015 · 8× Server Premier KSM64R52BD4-128MB 128GB DDR5-6400 RDIMM: cost per GB/sArithmetic on observed inputs only. No tunable assumption enters it. cost_per_gbs = landed_cost / memory_bandwidth_gbs = $52,403 / 307.2 GB/s = $170.58 per GB/s landed_cost 52403.32 USD — rig_landed = Σ part landed_total memory_bandwidth_gbs 307.2 GB/s — the published bandwidth of the rig's fastest memory pool Some part prices are older than 36 h. LOWER BOUND: 6 of 6 parts come from a retailer that does not state a delivery charge (SabrePC, Newegg, Newegg, ARCTIC, Newegg, Newegg), so the rig costs at least this and never less. Weights are read from the fastest pool first; a rig whose weights spill into a slower pool decodes below this figure.	$9,228 (estimate) ESTIMATECPU · EPYC 9015 · 8× Server Premier KSM64R52BD4-128MB 128GB DDR5-6400 RDIMM: cost per tok/sArithmetic that also depends on a tunable assumption, listed below. cost_per_tok_s = landed_cost / decode_tok_s = $52,403 / 5.68 tok/s = $9,228.43 per tok/s landed_cost 52403.32 USD — rig_landed = Σ part landed_total decode_tok_s 5.68 tok/s — GLM-5.3 decoding on this rig at the Plan's context Some part prices are older than 36 h. LOWER BOUND: 6 of 6 parts come from a retailer that does not state a delivery charge (SabrePC, Newegg, Newegg, ARCTIC, Newegg, Newegg), so the rig costs at least this and never less. MoE: only active parameters are read per token (SPEC §3). Effective bits include embeddings and higher-precision tensors in mixed quants. Assumes active weights are quantized like the file average. KV cache is re-read on every token (DECISIONS D-008). Batch size 1 (single user, interactive).
Priced from6 offers across 3 retailersoldeststale 44 h — past the 36 hour freshness windowopen the rig below for each part’s retailer, condition and route
12	Hybrid · RTX PRO 6000 Blackwell Max-Q Workstation Edition, 96GB + EPYC 9015 · 8× Server Premier KSM64R52BD4-96MC 96GB DDR5-6400 RDIMMDIY workstation · 7 parts · break down its parts	4 BLOCKERS	$47,638 (derived) DERIVEDHybrid · RTX PRO 6000 Blackwell Max-Q Workstation Edition, 96GB + EPYC 9015 · 8× Server Premier KSM64R52BD4-96MC 96GB DDR5-6400 RDIMM: landed costArithmetic on observed inputs only. No tunable assumption enters it. rig_landed = Σ part landed_total $638 + $895 + $32,414 + $12,721 + $75 + $192 + $704 = $47,638 AMD EPYC 9015 63809 ¢ — SabrePC · ship ASRock Rack GENOAD8QM3-2T/BCM 89458 ¢ — Newegg · ship Kingston Server Premier KSM64R52BD4-96MC 96GB DDR5-6400 RDIMM 3241392 ¢ — Newegg · ship NVIDIA RTX PRO 6000 Blackwell Max-Q Workstation Edition, 96GB 1272101 ¢ — SabrePC · ship ARCTIC Freezer 4U-SP5 7463 ¢ — ARCTIC · ship Phanteks Enthoo Pro 2 Server Edition 19191 ¢ — Newegg · ship Seasonic PRIME PX-2200 70371 ¢ — Newegg · ship Some part prices are older than 36 h. LOWER BOUND: 7 of 7 parts come from a retailer that does not state a delivery charge (SabrePC, Newegg, Newegg, SabrePC, ARCTIC, Newegg, Newegg), so the rig costs at least this and never less.	Unavailable — Purchase + 5-year power needs the power term. Load draw at the wall is unavailable: Electrical feasibility cannot be computed exactly: no load or TDP watts for ASRock Rack GENOAD8QM3-2T/BCM, Kingston Server Premier KSM64R52BD4-96MC 96GB DDR5-6400 RDIMM, ARCTIC Freezer 4U-SP5, Phanteks Enthoo Pro 2 Server Edition. (missing: ASRock Rack GENOAD8QM3-2T/BCM: load_watts or tdp_watts, Kingston Server Premier KSM64R52BD4-96MC 96GB DDR5-6400 RDIMM: load_watts or tdp_watts, ARCTIC Freezer 4U-SP5: load_watts or tdp_watts, Phanteks Enthoo Pro 2 Server Edition: load_watts or tdp_watts, AMD EPYC 9015: idle_watts, ASRock Rack GENOAD8QM3-2T/BCM: idle_watts, Kingston Server Premier KSM64R52BD4-96MC 96GB DDR5-6400 RDIMM: idle_watts, NVIDIA RTX PRO 6000 Blackwell Max-Q Workstation Edition, 96GB: idle_watts, ARCTIC Freezer 4U-SP5: idle_watts, Phanteks Enthoo Pro 2 Server Edition: idle_watts, Seasonic PRIME PX-2200: idle_watts)	$57.90 (derived) DERIVEDHybrid · RTX PRO 6000 Blackwell Max-Q Workstation Edition, 96GB + EPYC 9015 · 8× Server Premier KSM64R52BD4-96MC 96GB DDR5-6400 RDIMM: cost per GBArithmetic on observed inputs only. No tunable assumption enters it. cost_per_gb = landed_cost / usable_memory_gb = $47,638 / 822.72 GB = $57.9 /GB landed_cost 47637.85 USD — rig_landed = Σ part landed_total usable_memory_gb 822.72 GB — usable memory after the OS and framework reservation (Settings → Estimator) Some part prices are older than 36 h. LOWER BOUND: 7 of 7 parts come from a retailer that does not state a delivery charge (SabrePC, Newegg, Newegg, SabrePC, ARCTIC, Newegg, Newegg), so the rig costs at least this and never less. Usable memory is below physical memory for at least one pool (OS/driver reservation; editable in Settings).	$26.58 (derived) DERIVEDHybrid · RTX PRO 6000 Blackwell Max-Q Workstation Edition, 96GB + EPYC 9015 · 8× Server Premier KSM64R52BD4-96MC 96GB DDR5-6400 RDIMM: cost per GB/sArithmetic on observed inputs only. No tunable assumption enters it. cost_per_gbs = landed_cost / memory_bandwidth_gbs = $47,638 / 1,792 GB/s = $26.58 per GB/s landed_cost 47637.85 USD — rig_landed = Σ part landed_total memory_bandwidth_gbs 1792 GB/s — the published bandwidth of the rig's fastest memory pool Some part prices are older than 36 h. LOWER BOUND: 7 of 7 parts come from a retailer that does not state a delivery charge (SabrePC, Newegg, Newegg, SabrePC, ARCTIC, Newegg, Newegg), so the rig costs at least this and never less. Weights are read from the fastest pool first; a rig whose weights spill into a slower pool decodes below this figure.	$7,151 (estimate) ESTIMATEHybrid · RTX PRO 6000 Blackwell Max-Q Workstation Edition, 96GB + EPYC 9015 · 8× Server Premier KSM64R52BD4-96MC 96GB DDR5-6400 RDIMM: cost per tok/sArithmetic that also depends on a tunable assumption, listed below. cost_per_tok_s = landed_cost / decode_tok_s = $47,638 / 6.66 tok/s = $7,151.33 per tok/s landed_cost 47637.85 USD — rig_landed = Σ part landed_total decode_tok_s 6.66 tok/s — GLM-5.3 decoding on this rig at the Plan's context Some part prices are older than 36 h. LOWER BOUND: 7 of 7 parts come from a retailer that does not state a delivery charge (SabrePC, Newegg, Newegg, SabrePC, ARCTIC, Newegg, Newegg), so the rig costs at least this and never less. MoE: only active parameters are read per token (SPEC §3). Effective bits include embeddings and higher-precision tensors in mixed quants. Assumes active weights are quantized like the file average. KV cache is re-read on every token (DECISIONS D-008). Pools are read sequentially per token, so per-pool times add (harmonic combination). Per-token overhead on a split placement is DERIVED, not measured: each pool's own single-pool figure is charged in proportion to the share of each token it serves, because the fixed cost is per operation and the split puts that share of the operations on each device. Calibrated with that pool serving the whole model (D-013): GPU VRAM (1.58 ms), System RAM (44.61 ms). Batch size 1 (single user, interactive).
Priced from7 offers across 3 retailersoldeststale 44 h — past the 36 hour freshness windowopen the rig below for each part’s retailer, condition and route
13	3× RTX PRO 6000 Blackwell Max-Q Workstation Edition, 96GB · EPYC 9015 · 8× M321R4GA3BB6-CQK 32GB DDR5-4800 RDIMMDIY workstation · 7 parts · break down its parts	4 BLOCKERS	$42,494 (derived) DERIVED3× RTX PRO 6000 Blackwell Max-Q Workstation Edition, 96GB · EPYC 9015 · 8× M321R4GA3BB6-CQK 32GB DDR5-4800 RDIMM: landed costArithmetic on observed inputs only. No tunable assumption enters it. rig_landed = Σ part landed_total $638 + $895 + $1,828 + $38,163 + $75 + $192 + $704 = $42,494 AMD EPYC 9015 63809 ¢ — SabrePC · ship ASRock Rack GENOAD8QM3-2T/BCM 89458 ¢ — Newegg · ship Samsung M321R4GA3BB6-CQK 32GB DDR5-4800 RDIMM 182776 ¢ — SabrePC · ship NVIDIA RTX PRO 6000 Blackwell Max-Q Workstation Edition, 96GB 3816303 ¢ — SabrePC · ship ARCTIC Freezer 4U-SP5 7463 ¢ — ARCTIC · ship Phanteks Enthoo Pro 2 Server Edition 19191 ¢ — Newegg · ship Seasonic PRIME PX-2200 70371 ¢ — Newegg · ship Some part prices are older than 36 h. LOWER BOUND: 7 of 7 parts come from a retailer that does not state a delivery charge (SabrePC, Newegg, SabrePC, SabrePC, ARCTIC, Newegg, Newegg), so the rig costs at least this and never less.	Unavailable — Purchase + 5-year power needs the power term. Load draw at the wall is unavailable: Electrical feasibility cannot be computed exactly: no load or TDP watts for ASRock Rack GENOAD8QM3-2T/BCM, Samsung M321R4GA3BB6-CQK 32GB DDR5-4800 RDIMM, ARCTIC Freezer 4U-SP5, Phanteks Enthoo Pro 2 Server Edition. (missing: ASRock Rack GENOAD8QM3-2T/BCM: load_watts or tdp_watts, Samsung M321R4GA3BB6-CQK 32GB DDR5-4800 RDIMM: load_watts or tdp_watts, ARCTIC Freezer 4U-SP5: load_watts or tdp_watts, Phanteks Enthoo Pro 2 Server Edition: load_watts or tdp_watts, AMD EPYC 9015: idle_watts, ASRock Rack GENOAD8QM3-2T/BCM: idle_watts, Samsung M321R4GA3BB6-CQK 32GB DDR5-4800 RDIMM: idle_watts, NVIDIA RTX PRO 6000 Blackwell Max-Q Workstation Edition, 96GB: idle_watts, ARCTIC Freezer 4U-SP5: idle_watts, Phanteks Enthoo Pro 2 Server Edition: idle_watts, Seasonic PRIME PX-2200: idle_watts)	$81.32 (derived) DERIVED3× RTX PRO 6000 Blackwell Max-Q Workstation Edition, 96GB · EPYC 9015 · 8× M321R4GA3BB6-CQK 32GB DDR5-4800 RDIMM: cost per GBArithmetic on observed inputs only. No tunable assumption enters it. cost_per_gb = landed_cost / usable_memory_gb = $42,494 / 522.56 GB = $81.32 /GB landed_cost 42493.71 USD — rig_landed = Σ part landed_total usable_memory_gb 522.56 GB — usable memory after the OS and framework reservation (Settings → Estimator) Some part prices are older than 36 h. LOWER BOUND: 7 of 7 parts come from a retailer that does not state a delivery charge (SabrePC, Newegg, SabrePC, SabrePC, ARCTIC, Newegg, Newegg), so the rig costs at least this and never less. Usable memory is below physical memory for at least one pool (OS/driver reservation; editable in Settings).	$23.71 (derived) DERIVED3× RTX PRO 6000 Blackwell Max-Q Workstation Edition, 96GB · EPYC 9015 · 8× M321R4GA3BB6-CQK 32GB DDR5-4800 RDIMM: cost per GB/sArithmetic on observed inputs only. No tunable assumption enters it. cost_per_gbs = landed_cost / memory_bandwidth_gbs = $42,494 / 1,792 GB/s = $23.71 per GB/s landed_cost 42493.71 USD — rig_landed = Σ part landed_total memory_bandwidth_gbs 1792 GB/s — the published bandwidth of the rig's fastest memory pool Some part prices are older than 36 h. LOWER BOUND: 7 of 7 parts come from a retailer that does not state a delivery charge (SabrePC, Newegg, SabrePC, SabrePC, ARCTIC, Newegg, Newegg), so the rig costs at least this and never less. Weights are read from the fastest pool first; a rig whose weights spill into a slower pool decodes below this figure.	$4,171 (estimate) ESTIMATE3× RTX PRO 6000 Blackwell Max-Q Workstation Edition, 96GB · EPYC 9015 · 8× M321R4GA3BB6-CQK 32GB DDR5-4800 RDIMM: cost per tok/sArithmetic that also depends on a tunable assumption, listed below. cost_per_tok_s = landed_cost / decode_tok_s = $42,494 / 10.19 tok/s = $4,170.69 per tok/s landed_cost 42493.71 USD — rig_landed = Σ part landed_total decode_tok_s 10.19 tok/s — GLM-5.3 decoding on this rig at the Plan's context Some part prices are older than 36 h. LOWER BOUND: 7 of 7 parts come from a retailer that does not state a delivery charge (SabrePC, Newegg, SabrePC, SabrePC, ARCTIC, Newegg, Newegg), so the rig costs at least this and never less. MoE: only active parameters are read per token (SPEC §3). Effective bits include embeddings and higher-precision tensors in mixed quants. Assumes active weights are quantized like the file average. KV cache is re-read on every token (DECISIONS D-008). Pools are read sequentially per token, so per-pool times add (harmonic combination). layer split across 3 GPUs: capacity adds, per-token bandwidth does not (layers run sequentially) Per-token overhead on a split placement is DERIVED, not measured: each pool's own single-pool figure is charged in proportion to the share of each token it serves, because the fixed cost is per operation and the split puts that share of the operations on each device. Calibrated with that pool serving the whole model (D-013): GPU VRAM (1.58 ms), System RAM (44.61 ms). Batch size 1 (single user, interactive).
Priced from7 offers across 3 retailersoldeststale 44 h — past the 36 hour freshness windowopen the rig below for each part’s retailer, condition and route
14	Hybrid · GeForce RTX 5090 32G GAMING TRIO OC + EPYC 9015 · 8× Server Premier KSM64R52BD4-96MC 96GB DDR5-6400 RDIMMDIY workstation · 7 parts · break down its parts	4 BLOCKERS	$40,355 (derived) DERIVEDHybrid · GeForce RTX 5090 32G GAMING TRIO OC + EPYC 9015 · 8× Server Premier KSM64R52BD4-96MC 96GB DDR5-6400 RDIMM: landed costArithmetic on observed inputs only. No tunable assumption enters it. rig_landed = Σ part landed_total $638 + $895 + $32,414 + $5,438 + $75 + $192 + $704 = $40,355 AMD EPYC 9015 63809 ¢ — SabrePC · ship ASRock Rack GENOAD8QM3-2T/BCM 89458 ¢ — Newegg · ship Kingston Server Premier KSM64R52BD4-96MC 96GB DDR5-6400 RDIMM 3241392 ¢ — Newegg · ship MSI GeForce RTX 5090 32G GAMING TRIO OC 543786 ¢ — Newegg · ship ARCTIC Freezer 4U-SP5 7463 ¢ — ARCTIC · ship Phanteks Enthoo Pro 2 Server Edition 19191 ¢ — Newegg · ship Seasonic PRIME PX-2200 70371 ¢ — Newegg · ship Some part prices are older than 36 h. LOWER BOUND: 7 of 7 parts come from a retailer that does not state a delivery charge (SabrePC, Newegg, Newegg, Newegg, ARCTIC, Newegg, Newegg), so the rig costs at least this and never less.	Unavailable — Purchase + 5-year power needs the power term. Load draw at the wall is unavailable: Electrical feasibility cannot be computed exactly: no load or TDP watts for ASRock Rack GENOAD8QM3-2T/BCM, Kingston Server Premier KSM64R52BD4-96MC 96GB DDR5-6400 RDIMM, ARCTIC Freezer 4U-SP5, Phanteks Enthoo Pro 2 Server Edition. (missing: ASRock Rack GENOAD8QM3-2T/BCM: load_watts or tdp_watts, Kingston Server Premier KSM64R52BD4-96MC 96GB DDR5-6400 RDIMM: load_watts or tdp_watts, ARCTIC Freezer 4U-SP5: load_watts or tdp_watts, Phanteks Enthoo Pro 2 Server Edition: load_watts or tdp_watts, AMD EPYC 9015: idle_watts, ASRock Rack GENOAD8QM3-2T/BCM: idle_watts, Kingston Server Premier KSM64R52BD4-96MC 96GB DDR5-6400 RDIMM: idle_watts, MSI GeForce RTX 5090 32G GAMING TRIO OC: idle_watts, ARCTIC Freezer 4U-SP5: idle_watts, Phanteks Enthoo Pro 2 Server Edition: idle_watts, Seasonic PRIME PX-2200: idle_watts)	$53.05 (derived) DERIVEDHybrid · GeForce RTX 5090 32G GAMING TRIO OC + EPYC 9015 · 8× Server Premier KSM64R52BD4-96MC 96GB DDR5-6400 RDIMM: cost per GBArithmetic on observed inputs only. No tunable assumption enters it. cost_per_gb = landed_cost / usable_memory_gb = $40,355 / 760.64 GB = $53.05 /GB landed_cost 40354.7 USD — rig_landed = Σ part landed_total usable_memory_gb 760.64 GB — usable memory after the OS and framework reservation (Settings → Estimator) Some part prices are older than 36 h. LOWER BOUND: 7 of 7 parts come from a retailer that does not state a delivery charge (SabrePC, Newegg, Newegg, Newegg, ARCTIC, Newegg, Newegg), so the rig costs at least this and never less. Usable memory is below physical memory for at least one pool (OS/driver reservation; editable in Settings).	$22.52 (derived) DERIVEDHybrid · GeForce RTX 5090 32G GAMING TRIO OC + EPYC 9015 · 8× Server Premier KSM64R52BD4-96MC 96GB DDR5-6400 RDIMM: cost per GB/sArithmetic on observed inputs only. No tunable assumption enters it. cost_per_gbs = landed_cost / memory_bandwidth_gbs = $40,355 / 1,792 GB/s = $22.52 per GB/s landed_cost 40354.7 USD — rig_landed = Σ part landed_total memory_bandwidth_gbs 1792 GB/s — the published bandwidth of the rig's fastest memory pool Some part prices are older than 36 h. LOWER BOUND: 7 of 7 parts come from a retailer that does not state a delivery charge (SabrePC, Newegg, Newegg, Newegg, ARCTIC, Newegg, Newegg), so the rig costs at least this and never less. Weights are read from the fastest pool first; a rig whose weights spill into a slower pool decodes below this figure.	$6,757 (estimate) ESTIMATEHybrid · GeForce RTX 5090 32G GAMING TRIO OC + EPYC 9015 · 8× Server Premier KSM64R52BD4-96MC 96GB DDR5-6400 RDIMM: cost per tok/sArithmetic that also depends on a tunable assumption, listed below. cost_per_tok_s = landed_cost / decode_tok_s = $40,355 / 5.97 tok/s = $6,757.08 per tok/s landed_cost 40354.7 USD — rig_landed = Σ part landed_total decode_tok_s 5.97 tok/s — GLM-5.3 decoding on this rig at the Plan's context Some part prices are older than 36 h. LOWER BOUND: 7 of 7 parts come from a retailer that does not state a delivery charge (SabrePC, Newegg, Newegg, Newegg, ARCTIC, Newegg, Newegg), so the rig costs at least this and never less. MoE: only active parameters are read per token (SPEC §3). Effective bits include embeddings and higher-precision tensors in mixed quants. Assumes active weights are quantized like the file average. KV cache is re-read on every token (DECISIONS D-008). Pools are read sequentially per token, so per-pool times add (harmonic combination). Per-token overhead on a split placement is DERIVED, not measured: each pool's own single-pool figure is charged in proportion to the share of each token it serves, because the fixed cost is per operation and the split puts that share of the operations on each device. Calibrated with that pool serving the whole model (D-013): GPU VRAM (1.58 ms), System RAM (44.61 ms). Batch size 1 (single user, interactive).
Priced from7 offers across 3 retailersoldeststale 44 h — past the 36 hour freshness windowopen the rig below for each part’s retailer, condition and route
15	Hybrid · Radeon AI PRO R9700 32GB (32358-01-20G) + EPYC 9015 · 8× Server Premier KSM64R52BD4-96MC 96GB DDR5-6400 RDIMMDIY workstation · 7 parts · break down its parts	5 BLOCKERS	$36,836 (derived) DERIVEDHybrid · Radeon AI PRO R9700 32GB (32358-01-20G) + EPYC 9015 · 8× Server Premier KSM64R52BD4-96MC 96GB DDR5-6400 RDIMM: landed costArithmetic on observed inputs only. No tunable assumption enters it. rig_landed = Σ part landed_total $638 + $895 + $32,414 + $1,919 + $75 + $192 + $704 = $36,836 AMD EPYC 9015 63809 ¢ — SabrePC · ship ASRock Rack GENOAD8QM3-2T/BCM 89458 ¢ — Newegg · ship Kingston Server Premier KSM64R52BD4-96MC 96GB DDR5-6400 RDIMM 3241392 ¢ — Newegg · ship Sapphire Radeon AI PRO R9700 32GB (32358-01-20G) 191924 ¢ — Newegg · ship ARCTIC Freezer 4U-SP5 7463 ¢ — ARCTIC · ship Phanteks Enthoo Pro 2 Server Edition 19191 ¢ — Newegg · ship Seasonic PRIME PX-2200 70371 ¢ — Newegg · ship Some part prices are older than 36 h. LOWER BOUND: 7 of 7 parts come from a retailer that does not state a delivery charge (SabrePC, Newegg, Newegg, Newegg, ARCTIC, Newegg, Newegg), so the rig costs at least this and never less.	Unavailable — Purchase + 5-year power needs the power term. Load draw at the wall is unavailable: Electrical feasibility cannot be computed exactly: no load or TDP watts for ASRock Rack GENOAD8QM3-2T/BCM, Kingston Server Premier KSM64R52BD4-96MC 96GB DDR5-6400 RDIMM, ARCTIC Freezer 4U-SP5, Phanteks Enthoo Pro 2 Server Edition. (missing: ASRock Rack GENOAD8QM3-2T/BCM: load_watts or tdp_watts, Kingston Server Premier KSM64R52BD4-96MC 96GB DDR5-6400 RDIMM: load_watts or tdp_watts, ARCTIC Freezer 4U-SP5: load_watts or tdp_watts, Phanteks Enthoo Pro 2 Server Edition: load_watts or tdp_watts, AMD EPYC 9015: idle_watts, ASRock Rack GENOAD8QM3-2T/BCM: idle_watts, Kingston Server Premier KSM64R52BD4-96MC 96GB DDR5-6400 RDIMM: idle_watts, Sapphire Radeon AI PRO R9700 32GB (32358-01-20G): idle_watts, ARCTIC Freezer 4U-SP5: idle_watts, Phanteks Enthoo Pro 2 Server Edition: idle_watts, Seasonic PRIME PX-2200: idle_watts)	$48.43 (derived) DERIVEDHybrid · Radeon AI PRO R9700 32GB (32358-01-20G) + EPYC 9015 · 8× Server Premier KSM64R52BD4-96MC 96GB DDR5-6400 RDIMM: cost per GBArithmetic on observed inputs only. No tunable assumption enters it. cost_per_gb = landed_cost / usable_memory_gb = $36,836 / 760.64 GB = $48.43 /GB landed_cost 36836.08 USD — rig_landed = Σ part landed_total usable_memory_gb 760.64 GB — usable memory after the OS and framework reservation (Settings → Estimator) Some part prices are older than 36 h. LOWER BOUND: 7 of 7 parts come from a retailer that does not state a delivery charge (SabrePC, Newegg, Newegg, Newegg, ARCTIC, Newegg, Newegg), so the rig costs at least this and never less. Usable memory is below physical memory for at least one pool (OS/driver reservation; editable in Settings).	$57.56 (derived) DERIVEDHybrid · Radeon AI PRO R9700 32GB (32358-01-20G) + EPYC 9015 · 8× Server Premier KSM64R52BD4-96MC 96GB DDR5-6400 RDIMM: cost per GB/sArithmetic on observed inputs only. No tunable assumption enters it. cost_per_gbs = landed_cost / memory_bandwidth_gbs = $36,836 / 640 GB/s = $57.56 per GB/s landed_cost 36836.08 USD — rig_landed = Σ part landed_total memory_bandwidth_gbs 640 GB/s — the published bandwidth of the rig's fastest memory pool Some part prices are older than 36 h. LOWER BOUND: 7 of 7 parts come from a retailer that does not state a delivery charge (SabrePC, Newegg, Newegg, Newegg, ARCTIC, Newegg, Newegg), so the rig costs at least this and never less. Weights are read from the fastest pool first; a rig whose weights spill into a slower pool decodes below this figure.	$6,301 (estimate) ESTIMATEHybrid · Radeon AI PRO R9700 32GB (32358-01-20G) + EPYC 9015 · 8× Server Premier KSM64R52BD4-96MC 96GB DDR5-6400 RDIMM: cost per tok/sArithmetic that also depends on a tunable assumption, listed below. cost_per_tok_s = landed_cost / decode_tok_s = $36,836 / 5.85 tok/s = $6,301.2 per tok/s landed_cost 36836.08 USD — rig_landed = Σ part landed_total decode_tok_s 5.85 tok/s — GLM-5.3 decoding on this rig at the Plan's context Some part prices are older than 36 h. LOWER BOUND: 7 of 7 parts come from a retailer that does not state a delivery charge (SabrePC, Newegg, Newegg, Newegg, ARCTIC, Newegg, Newegg), so the rig costs at least this and never less. MoE: only active parameters are read per token (SPEC §3). Effective bits include embeddings and higher-precision tensors in mixed quants. Assumes active weights are quantized like the file average. KV cache is re-read on every token (DECISIONS D-008). Pools are read sequentially per token, so per-pool times add (harmonic combination). Per-token overhead on a split placement is DERIVED, not measured: each pool's own single-pool figure is charged in proportion to the share of each token it serves, because the fixed cost is per operation and the split puts that share of the operations on each device. Calibrated with that pool serving the whole model (D-013): GPU VRAM (1.58 ms), System RAM (44.61 ms). Batch size 1 (single user, interactive).
Priced from7 offers across 3 retailersoldeststale 44 h — past the 36 hour freshness windowopen the rig below for each part’s retailer, condition and route
16	Hybrid · AI TOP Radeon AI PRO R9700 32GB (GV-R9700AI TOP-32GD) + EPYC 9015 · 8× Server Premier KSM64R52BD4-96MC 96GB DDR5-6400 RDIMMDIY workstation · 7 parts · break down its parts	5 BLOCKERS	$36,741 (derived) DERIVEDHybrid · AI TOP Radeon AI PRO R9700 32GB (GV-R9700AI TOP-32GD) + EPYC 9015 · 8× Server Premier KSM64R52BD4-96MC 96GB DDR5-6400 RDIMM: landed costArithmetic on observed inputs only. No tunable assumption enters it. rig_landed = Σ part landed_total $638 + $895 + $32,414 + $1,824 + $75 + $192 + $704 = $36,741 AMD EPYC 9015 63809 ¢ — SabrePC · ship ASRock Rack GENOAD8QM3-2T/BCM 89458 ¢ — Newegg · ship Kingston Server Premier KSM64R52BD4-96MC 96GB DDR5-6400 RDIMM 3241392 ¢ — Newegg · ship GIGABYTE AI TOP Radeon AI PRO R9700 32GB (GV-R9700AI TOP-32GD) 182424 ¢ — SabrePC · ship ARCTIC Freezer 4U-SP5 7463 ¢ — ARCTIC · ship Phanteks Enthoo Pro 2 Server Edition 19191 ¢ — Newegg · ship Seasonic PRIME PX-2200 70371 ¢ — Newegg · ship Some part prices are older than 36 h. LOWER BOUND: 7 of 7 parts come from a retailer that does not state a delivery charge (SabrePC, Newegg, Newegg, SabrePC, ARCTIC, Newegg, Newegg), so the rig costs at least this and never less.	Unavailable — Purchase + 5-year power needs the power term. Load draw at the wall is unavailable: Electrical feasibility cannot be computed exactly: no load or TDP watts for ASRock Rack GENOAD8QM3-2T/BCM, Kingston Server Premier KSM64R52BD4-96MC 96GB DDR5-6400 RDIMM, ARCTIC Freezer 4U-SP5, Phanteks Enthoo Pro 2 Server Edition. (missing: ASRock Rack GENOAD8QM3-2T/BCM: load_watts or tdp_watts, Kingston Server Premier KSM64R52BD4-96MC 96GB DDR5-6400 RDIMM: load_watts or tdp_watts, ARCTIC Freezer 4U-SP5: load_watts or tdp_watts, Phanteks Enthoo Pro 2 Server Edition: load_watts or tdp_watts, AMD EPYC 9015: idle_watts, ASRock Rack GENOAD8QM3-2T/BCM: idle_watts, Kingston Server Premier KSM64R52BD4-96MC 96GB DDR5-6400 RDIMM: idle_watts, GIGABYTE AI TOP Radeon AI PRO R9700 32GB (GV-R9700AI TOP-32GD): idle_watts, ARCTIC Freezer 4U-SP5: idle_watts, Phanteks Enthoo Pro 2 Server Edition: idle_watts, Seasonic PRIME PX-2200: idle_watts)	$48.30 (derived) DERIVEDHybrid · AI TOP Radeon AI PRO R9700 32GB (GV-R9700AI TOP-32GD) + EPYC 9015 · 8× Server Premier KSM64R52BD4-96MC 96GB DDR5-6400 RDIMM: cost per GBArithmetic on observed inputs only. No tunable assumption enters it. cost_per_gb = landed_cost / usable_memory_gb = $36,741 / 760.64 GB = $48.3 /GB landed_cost 36741.08 USD — rig_landed = Σ part landed_total usable_memory_gb 760.64 GB — usable memory after the OS and framework reservation (Settings → Estimator) Some part prices are older than 36 h. LOWER BOUND: 7 of 7 parts come from a retailer that does not state a delivery charge (SabrePC, Newegg, Newegg, SabrePC, ARCTIC, Newegg, Newegg), so the rig costs at least this and never less. Usable memory is below physical memory for at least one pool (OS/driver reservation; editable in Settings).	$57.41 (derived) DERIVEDHybrid · AI TOP Radeon AI PRO R9700 32GB (GV-R9700AI TOP-32GD) + EPYC 9015 · 8× Server Premier KSM64R52BD4-96MC 96GB DDR5-6400 RDIMM: cost per GB/sArithmetic on observed inputs only. No tunable assumption enters it. cost_per_gbs = landed_cost / memory_bandwidth_gbs = $36,741 / 640 GB/s = $57.41 per GB/s landed_cost 36741.08 USD — rig_landed = Σ part landed_total memory_bandwidth_gbs 640 GB/s — the published bandwidth of the rig's fastest memory pool Some part prices are older than 36 h. LOWER BOUND: 7 of 7 parts come from a retailer that does not state a delivery charge (SabrePC, Newegg, Newegg, SabrePC, ARCTIC, Newegg, Newegg), so the rig costs at least this and never less. Weights are read from the fastest pool first; a rig whose weights spill into a slower pool decodes below this figure.	$6,285 (estimate) ESTIMATEHybrid · AI TOP Radeon AI PRO R9700 32GB (GV-R9700AI TOP-32GD) + EPYC 9015 · 8× Server Premier KSM64R52BD4-96MC 96GB DDR5-6400 RDIMM: cost per tok/sArithmetic that also depends on a tunable assumption, listed below. cost_per_tok_s = landed_cost / decode_tok_s = $36,741 / 5.85 tok/s = $6,284.95 per tok/s landed_cost 36741.08 USD — rig_landed = Σ part landed_total decode_tok_s 5.85 tok/s — GLM-5.3 decoding on this rig at the Plan's context Some part prices are older than 36 h. LOWER BOUND: 7 of 7 parts come from a retailer that does not state a delivery charge (SabrePC, Newegg, Newegg, SabrePC, ARCTIC, Newegg, Newegg), so the rig costs at least this and never less. MoE: only active parameters are read per token (SPEC §3). Effective bits include embeddings and higher-precision tensors in mixed quants. Assumes active weights are quantized like the file average. KV cache is re-read on every token (DECISIONS D-008). Pools are read sequentially per token, so per-pool times add (harmonic combination). Per-token overhead on a split placement is DERIVED, not measured: each pool's own single-pool figure is charged in proportion to the share of each token it serves, because the fixed cost is per operation and the split puts that share of the operations on each device. Calibrated with that pool serving the whole model (D-013): GPU VRAM (1.58 ms), System RAM (44.61 ms). Batch size 1 (single user, interactive).
Priced from7 offers across 3 retailersoldeststale 44 h — past the 36 hour freshness windowopen the rig below for each part’s retailer, condition and route
17	Hybrid · GeForce RTX 5080 WINDFORCE OC 16G (GV-N5080WF3OC-16GD) + EPYC 9015 · 8× Server Premier KSM64R52BD4-96MC 96GB DDR5-6400 RDIMMDIY workstation · 7 parts · break down its parts	4 BLOCKERS	$36,729 (derived) DERIVEDHybrid · GeForce RTX 5080 WINDFORCE OC 16G (GV-N5080WF3OC-16GD) + EPYC 9015 · 8× Server Premier KSM64R52BD4-96MC 96GB DDR5-6400 RDIMM: landed costArithmetic on observed inputs only. No tunable assumption enters it. rig_landed = Σ part landed_total $638 + $895 + $32,414 + $1,813 + $75 + $192 + $704 = $36,729 AMD EPYC 9015 63809 ¢ — SabrePC · ship ASRock Rack GENOAD8QM3-2T/BCM 89458 ¢ — Newegg · ship Kingston Server Premier KSM64R52BD4-96MC 96GB DDR5-6400 RDIMM 3241392 ¢ — Newegg · ship GIGABYTE GeForce RTX 5080 WINDFORCE OC 16G (GV-N5080WF3OC-16GD) 181261 ¢ — Newegg · ship ARCTIC Freezer 4U-SP5 7463 ¢ — ARCTIC · ship Phanteks Enthoo Pro 2 Server Edition 19191 ¢ — Newegg · ship Seasonic PRIME PX-2200 70371 ¢ — Newegg · ship Some part prices are older than 36 h. LOWER BOUND: 7 of 7 parts come from a retailer that does not state a delivery charge (SabrePC, Newegg, Newegg, Newegg, ARCTIC, Newegg, Newegg), so the rig costs at least this and never less.	Unavailable — Purchase + 5-year power needs the power term. Load draw at the wall is unavailable: Electrical feasibility cannot be computed exactly: no load or TDP watts for ASRock Rack GENOAD8QM3-2T/BCM, Kingston Server Premier KSM64R52BD4-96MC 96GB DDR5-6400 RDIMM, ARCTIC Freezer 4U-SP5, Phanteks Enthoo Pro 2 Server Edition. (missing: ASRock Rack GENOAD8QM3-2T/BCM: load_watts or tdp_watts, Kingston Server Premier KSM64R52BD4-96MC 96GB DDR5-6400 RDIMM: load_watts or tdp_watts, ARCTIC Freezer 4U-SP5: load_watts or tdp_watts, Phanteks Enthoo Pro 2 Server Edition: load_watts or tdp_watts, AMD EPYC 9015: idle_watts, ASRock Rack GENOAD8QM3-2T/BCM: idle_watts, Kingston Server Premier KSM64R52BD4-96MC 96GB DDR5-6400 RDIMM: idle_watts, GIGABYTE GeForce RTX 5080 WINDFORCE OC 16G (GV-N5080WF3OC-16GD): idle_watts, ARCTIC Freezer 4U-SP5: idle_watts, Phanteks Enthoo Pro 2 Server Edition: idle_watts, Seasonic PRIME PX-2200: idle_watts)	$49.29 (derived) DERIVEDHybrid · GeForce RTX 5080 WINDFORCE OC 16G (GV-N5080WF3OC-16GD) + EPYC 9015 · 8× Server Premier KSM64R52BD4-96MC 96GB DDR5-6400 RDIMM: cost per GBArithmetic on observed inputs only. No tunable assumption enters it. cost_per_gb = landed_cost / usable_memory_gb = $36,729 / 745.12 GB = $49.29 /GB landed_cost 36729.45 USD — rig_landed = Σ part landed_total usable_memory_gb 745.12 GB — usable memory after the OS and framework reservation (Settings → Estimator) Some part prices are older than 36 h. LOWER BOUND: 7 of 7 parts come from a retailer that does not state a delivery charge (SabrePC, Newegg, Newegg, Newegg, ARCTIC, Newegg, Newegg), so the rig costs at least this and never less. Usable memory is below physical memory for at least one pool (OS/driver reservation; editable in Settings).	$38.26 (derived) DERIVEDHybrid · GeForce RTX 5080 WINDFORCE OC 16G (GV-N5080WF3OC-16GD) + EPYC 9015 · 8× Server Premier KSM64R52BD4-96MC 96GB DDR5-6400 RDIMM: cost per GB/sArithmetic on observed inputs only. No tunable assumption enters it. cost_per_gbs = landed_cost / memory_bandwidth_gbs = $36,729 / 960 GB/s = $38.26 per GB/s landed_cost 36729.45 USD — rig_landed = Σ part landed_total memory_bandwidth_gbs 960 GB/s — the published bandwidth of the rig's fastest memory pool Some part prices are older than 36 h. LOWER BOUND: 7 of 7 parts come from a retailer that does not state a delivery charge (SabrePC, Newegg, Newegg, Newegg, ARCTIC, Newegg, Newegg), so the rig costs at least this and never less. Weights are read from the fastest pool first; a rig whose weights spill into a slower pool decodes below this figure.	$6,341 (estimate) ESTIMATEHybrid · GeForce RTX 5080 WINDFORCE OC 16G (GV-N5080WF3OC-16GD) + EPYC 9015 · 8× Server Premier KSM64R52BD4-96MC 96GB DDR5-6400 RDIMM: cost per tok/sArithmetic that also depends on a tunable assumption, listed below. cost_per_tok_s = landed_cost / decode_tok_s = $36,729 / 5.79 tok/s = $6,341.12 per tok/s landed_cost 36729.45 USD — rig_landed = Σ part landed_total decode_tok_s 5.79 tok/s — GLM-5.3 decoding on this rig at the Plan's context Some part prices are older than 36 h. LOWER BOUND: 7 of 7 parts come from a retailer that does not state a delivery charge (SabrePC, Newegg, Newegg, Newegg, ARCTIC, Newegg, Newegg), so the rig costs at least this and never less. MoE: only active parameters are read per token (SPEC §3). Effective bits include embeddings and higher-precision tensors in mixed quants. Assumes active weights are quantized like the file average. KV cache is re-read on every token (DECISIONS D-008). Pools are read sequentially per token, so per-pool times add (harmonic combination). Per-token overhead on a split placement is DERIVED, not measured: each pool's own single-pool figure is charged in proportion to the share of each token it serves, because the fixed cost is per operation and the split puts that share of the operations on each device. Calibrated with that pool serving the whole model (D-013): GPU VRAM (1.58 ms), System RAM (44.61 ms). Batch size 1 (single user, interactive).
Priced from7 offers across 3 retailersoldeststale 44 h — past the 36 hour freshness windowopen the rig below for each part’s retailer, condition and route
18	Hybrid · Radeon AI PRO R9700 Creator 32GB (R9700 CT) + EPYC 9015 · 8× Server Premier KSM64R52BD4-96MC 96GB DDR5-6400 RDIMMDIY workstation · 7 parts · break down its parts	5 BLOCKERS	$36,729 (derived) DERIVEDHybrid · Radeon AI PRO R9700 Creator 32GB (R9700 CT) + EPYC 9015 · 8× Server Premier KSM64R52BD4-96MC 96GB DDR5-6400 RDIMM: landed costArithmetic on observed inputs only. No tunable assumption enters it. rig_landed = Σ part landed_total $638 + $895 + $32,414 + $1,813 + $75 + $192 + $704 = $36,729 AMD EPYC 9015 63809 ¢ — SabrePC · ship ASRock Rack GENOAD8QM3-2T/BCM 89458 ¢ — Newegg · ship Kingston Server Premier KSM64R52BD4-96MC 96GB DDR5-6400 RDIMM 3241392 ¢ — Newegg · ship ASRock Radeon AI PRO R9700 Creator 32GB (R9700 CT) 181261 ¢ — Newegg · ship ARCTIC Freezer 4U-SP5 7463 ¢ — ARCTIC · ship Phanteks Enthoo Pro 2 Server Edition 19191 ¢ — Newegg · ship Seasonic PRIME PX-2200 70371 ¢ — Newegg · ship Some part prices are older than 36 h. LOWER BOUND: 7 of 7 parts come from a retailer that does not state a delivery charge (SabrePC, Newegg, Newegg, Newegg, ARCTIC, Newegg, Newegg), so the rig costs at least this and never less.	Unavailable — Purchase + 5-year power needs the power term. Load draw at the wall is unavailable: Electrical feasibility cannot be computed exactly: no load or TDP watts for ASRock Rack GENOAD8QM3-2T/BCM, Kingston Server Premier KSM64R52BD4-96MC 96GB DDR5-6400 RDIMM, ARCTIC Freezer 4U-SP5, Phanteks Enthoo Pro 2 Server Edition. (missing: ASRock Rack GENOAD8QM3-2T/BCM: load_watts or tdp_watts, Kingston Server Premier KSM64R52BD4-96MC 96GB DDR5-6400 RDIMM: load_watts or tdp_watts, ARCTIC Freezer 4U-SP5: load_watts or tdp_watts, Phanteks Enthoo Pro 2 Server Edition: load_watts or tdp_watts, AMD EPYC 9015: idle_watts, ASRock Rack GENOAD8QM3-2T/BCM: idle_watts, Kingston Server Premier KSM64R52BD4-96MC 96GB DDR5-6400 RDIMM: idle_watts, ASRock Radeon AI PRO R9700 Creator 32GB (R9700 CT): idle_watts, ARCTIC Freezer 4U-SP5: idle_watts, Phanteks Enthoo Pro 2 Server Edition: idle_watts, Seasonic PRIME PX-2200: idle_watts)	$48.29 (derived) DERIVEDHybrid · Radeon AI PRO R9700 Creator 32GB (R9700 CT) + EPYC 9015 · 8× Server Premier KSM64R52BD4-96MC 96GB DDR5-6400 RDIMM: cost per GBArithmetic on observed inputs only. No tunable assumption enters it. cost_per_gb = landed_cost / usable_memory_gb = $36,729 / 760.64 GB = $48.29 /GB landed_cost 36729.45 USD — rig_landed = Σ part landed_total usable_memory_gb 760.64 GB — usable memory after the OS and framework reservation (Settings → Estimator) Some part prices are older than 36 h. LOWER BOUND: 7 of 7 parts come from a retailer that does not state a delivery charge (SabrePC, Newegg, Newegg, Newegg, ARCTIC, Newegg, Newegg), so the rig costs at least this and never less. Usable memory is below physical memory for at least one pool (OS/driver reservation; editable in Settings).	$57.39 (derived) DERIVEDHybrid · Radeon AI PRO R9700 Creator 32GB (R9700 CT) + EPYC 9015 · 8× Server Premier KSM64R52BD4-96MC 96GB DDR5-6400 RDIMM: cost per GB/sArithmetic on observed inputs only. No tunable assumption enters it. cost_per_gbs = landed_cost / memory_bandwidth_gbs = $36,729 / 640 GB/s = $57.39 per GB/s landed_cost 36729.45 USD — rig_landed = Σ part landed_total memory_bandwidth_gbs 640 GB/s — the published bandwidth of the rig's fastest memory pool Some part prices are older than 36 h. LOWER BOUND: 7 of 7 parts come from a retailer that does not state a delivery charge (SabrePC, Newegg, Newegg, Newegg, ARCTIC, Newegg, Newegg), so the rig costs at least this and never less. Weights are read from the fastest pool first; a rig whose weights spill into a slower pool decodes below this figure.	$6,283 (estimate) ESTIMATEHybrid · Radeon AI PRO R9700 Creator 32GB (R9700 CT) + EPYC 9015 · 8× Server Premier KSM64R52BD4-96MC 96GB DDR5-6400 RDIMM: cost per tok/sArithmetic that also depends on a tunable assumption, listed below. cost_per_tok_s = landed_cost / decode_tok_s = $36,729 / 5.85 tok/s = $6,282.96 per tok/s landed_cost 36729.45 USD — rig_landed = Σ part landed_total decode_tok_s 5.85 tok/s — GLM-5.3 decoding on this rig at the Plan's context Some part prices are older than 36 h. LOWER BOUND: 7 of 7 parts come from a retailer that does not state a delivery charge (SabrePC, Newegg, Newegg, Newegg, ARCTIC, Newegg, Newegg), so the rig costs at least this and never less. MoE: only active parameters are read per token (SPEC §3). Effective bits include embeddings and higher-precision tensors in mixed quants. Assumes active weights are quantized like the file average. KV cache is re-read on every token (DECISIONS D-008). Pools are read sequentially per token, so per-pool times add (harmonic combination). Per-token overhead on a split placement is DERIVED, not measured: each pool's own single-pool figure is charged in proportion to the share of each token it serves, because the fixed cost is per operation and the split puts that share of the operations on each device. Calibrated with that pool serving the whole model (D-013): GPU VRAM (1.58 ms), System RAM (44.61 ms). Batch size 1 (single user, interactive).
Priced from7 offers across 3 retailersoldeststale 30 d — past the 36 hour freshness windowopen the rig below for each part’s retailer, condition and route
19	Hybrid · GeForce RTX 5080 16G SHADOW 3X OC + EPYC 9015 · 8× Server Premier KSM64R52BD4-96MC 96GB DDR5-6400 RDIMMDIY workstation · 7 parts · break down its parts	4 BLOCKERS	$36,719 (derived) DERIVEDHybrid · GeForce RTX 5080 16G SHADOW 3X OC + EPYC 9015 · 8× Server Premier KSM64R52BD4-96MC 96GB DDR5-6400 RDIMM: landed costArithmetic on observed inputs only. No tunable assumption enters it. rig_landed = Σ part landed_total $638 + $895 + $32,414 + $1,802 + $75 + $192 + $704 = $36,719 AMD EPYC 9015 63809 ¢ — SabrePC · ship ASRock Rack GENOAD8QM3-2T/BCM 89458 ¢ — Newegg · ship Kingston Server Premier KSM64R52BD4-96MC 96GB DDR5-6400 RDIMM 3241392 ¢ — Newegg · ship MSI GeForce RTX 5080 16G SHADOW 3X OC 180195 ¢ — Newegg · ship ARCTIC Freezer 4U-SP5 7463 ¢ — ARCTIC · ship Phanteks Enthoo Pro 2 Server Edition 19191 ¢ — Newegg · ship Seasonic PRIME PX-2200 70371 ¢ — Newegg · ship Some part prices are older than 36 h. LOWER BOUND: 7 of 7 parts come from a retailer that does not state a delivery charge (SabrePC, Newegg, Newegg, Newegg, ARCTIC, Newegg, Newegg), so the rig costs at least this and never less.	Unavailable — Purchase + 5-year power needs the power term. Load draw at the wall is unavailable: Electrical feasibility cannot be computed exactly: no load or TDP watts for ASRock Rack GENOAD8QM3-2T/BCM, Kingston Server Premier KSM64R52BD4-96MC 96GB DDR5-6400 RDIMM, ARCTIC Freezer 4U-SP5, Phanteks Enthoo Pro 2 Server Edition. (missing: ASRock Rack GENOAD8QM3-2T/BCM: load_watts or tdp_watts, Kingston Server Premier KSM64R52BD4-96MC 96GB DDR5-6400 RDIMM: load_watts or tdp_watts, ARCTIC Freezer 4U-SP5: load_watts or tdp_watts, Phanteks Enthoo Pro 2 Server Edition: load_watts or tdp_watts, AMD EPYC 9015: idle_watts, ASRock Rack GENOAD8QM3-2T/BCM: idle_watts, Kingston Server Premier KSM64R52BD4-96MC 96GB DDR5-6400 RDIMM: idle_watts, MSI GeForce RTX 5080 16G SHADOW 3X OC: idle_watts, ARCTIC Freezer 4U-SP5: idle_watts, Phanteks Enthoo Pro 2 Server Edition: idle_watts, Seasonic PRIME PX-2200: idle_watts)	$49.28 (derived) DERIVEDHybrid · GeForce RTX 5080 16G SHADOW 3X OC + EPYC 9015 · 8× Server Premier KSM64R52BD4-96MC 96GB DDR5-6400 RDIMM: cost per GBArithmetic on observed inputs only. No tunable assumption enters it. cost_per_gb = landed_cost / usable_memory_gb = $36,719 / 745.12 GB = $49.28 /GB landed_cost 36718.79 USD — rig_landed = Σ part landed_total usable_memory_gb 745.12 GB — usable memory after the OS and framework reservation (Settings → Estimator) Some part prices are older than 36 h. LOWER BOUND: 7 of 7 parts come from a retailer that does not state a delivery charge (SabrePC, Newegg, Newegg, Newegg, ARCTIC, Newegg, Newegg), so the rig costs at least this and never less. Usable memory is below physical memory for at least one pool (OS/driver reservation; editable in Settings).	$38.25 (derived) DERIVEDHybrid · GeForce RTX 5080 16G SHADOW 3X OC + EPYC 9015 · 8× Server Premier KSM64R52BD4-96MC 96GB DDR5-6400 RDIMM: cost per GB/sArithmetic on observed inputs only. No tunable assumption enters it. cost_per_gbs = landed_cost / memory_bandwidth_gbs = $36,719 / 960 GB/s = $38.25 per GB/s landed_cost 36718.79 USD — rig_landed = Σ part landed_total memory_bandwidth_gbs 960 GB/s — the published bandwidth of the rig's fastest memory pool Some part prices are older than 36 h. LOWER BOUND: 7 of 7 parts come from a retailer that does not state a delivery charge (SabrePC, Newegg, Newegg, Newegg, ARCTIC, Newegg, Newegg), so the rig costs at least this and never less. Weights are read from the fastest pool first; a rig whose weights spill into a slower pool decodes below this figure.	$6,339 (estimate) ESTIMATEHybrid · GeForce RTX 5080 16G SHADOW 3X OC + EPYC 9015 · 8× Server Premier KSM64R52BD4-96MC 96GB DDR5-6400 RDIMM: cost per tok/sArithmetic that also depends on a tunable assumption, listed below. cost_per_tok_s = landed_cost / decode_tok_s = $36,719 / 5.79 tok/s = $6,339.28 per tok/s landed_cost 36718.79 USD — rig_landed = Σ part landed_total decode_tok_s 5.79 tok/s — GLM-5.3 decoding on this rig at the Plan's context Some part prices are older than 36 h. LOWER BOUND: 7 of 7 parts come from a retailer that does not state a delivery charge (SabrePC, Newegg, Newegg, Newegg, ARCTIC, Newegg, Newegg), so the rig costs at least this and never less. MoE: only active parameters are read per token (SPEC §3). Effective bits include embeddings and higher-precision tensors in mixed quants. Assumes active weights are quantized like the file average. KV cache is re-read on every token (DECISIONS D-008). Pools are read sequentially per token, so per-pool times add (harmonic combination). Per-token overhead on a split placement is DERIVED, not measured: each pool's own single-pool figure is charged in proportion to the share of each token it serves, because the fixed cost is per operation and the split puts that share of the operations on each device. Calibrated with that pool serving the whole model (D-013): GPU VRAM (1.58 ms), System RAM (44.61 ms). Batch size 1 (single user, interactive).
Priced from7 offers across 3 retailersoldeststale 44 h — past the 36 hour freshness windowopen the rig below for each part’s retailer, condition and route
20	CPU · EPYC 9015 · 8× Server Premier KSM64R52BD4-96MC 96GB DDR5-6400 RDIMMDIY workstation · 6 parts · break down its parts	5 BLOCKERS	$34,917 (derived) DERIVEDCPU · EPYC 9015 · 8× Server Premier KSM64R52BD4-96MC 96GB DDR5-6400 RDIMM: landed costArithmetic on observed inputs only. No tunable assumption enters it. rig_landed = Σ part landed_total $638 + $895 + $32,414 + $75 + $192 + $704 = $34,917 AMD EPYC 9015 63809 ¢ — SabrePC · ship ASRock Rack GENOAD8QM3-2T/BCM 89458 ¢ — Newegg · ship Kingston Server Premier KSM64R52BD4-96MC 96GB DDR5-6400 RDIMM 3241392 ¢ — Newegg · ship ARCTIC Freezer 4U-SP5 7463 ¢ — ARCTIC · ship Phanteks Enthoo Pro 2 Server Edition 19191 ¢ — Newegg · ship Seasonic PRIME PX-2200 70371 ¢ — Newegg · ship Some part prices are older than 36 h. LOWER BOUND: 6 of 6 parts come from a retailer that does not state a delivery charge (SabrePC, Newegg, Newegg, ARCTIC, Newegg, Newegg), so the rig costs at least this and never less.	Unavailable — Purchase + 5-year power needs the power term. Load draw at the wall is unavailable: Electrical feasibility cannot be computed exactly: no load or TDP watts for ASRock Rack GENOAD8QM3-2T/BCM, Kingston Server Premier KSM64R52BD4-96MC 96GB DDR5-6400 RDIMM, ARCTIC Freezer 4U-SP5, Phanteks Enthoo Pro 2 Server Edition. (missing: ASRock Rack GENOAD8QM3-2T/BCM: load_watts or tdp_watts, Kingston Server Premier KSM64R52BD4-96MC 96GB DDR5-6400 RDIMM: load_watts or tdp_watts, ARCTIC Freezer 4U-SP5: load_watts or tdp_watts, Phanteks Enthoo Pro 2 Server Edition: load_watts or tdp_watts, AMD EPYC 9015: idle_watts, ASRock Rack GENOAD8QM3-2T/BCM: idle_watts, Kingston Server Premier KSM64R52BD4-96MC 96GB DDR5-6400 RDIMM: idle_watts, ARCTIC Freezer 4U-SP5: idle_watts, Phanteks Enthoo Pro 2 Server Edition: idle_watts, Seasonic PRIME PX-2200: idle_watts)	$47.86 (derived) DERIVEDCPU · EPYC 9015 · 8× Server Premier KSM64R52BD4-96MC 96GB DDR5-6400 RDIMM: cost per GBArithmetic on observed inputs only. No tunable assumption enters it. cost_per_gb = landed_cost / usable_memory_gb = $34,917 / 729.6 GB = $47.86 /GB landed_cost 34916.84 USD — rig_landed = Σ part landed_total usable_memory_gb 729.6 GB — usable memory after the OS and framework reservation (Settings → Estimator) Some part prices are older than 36 h. LOWER BOUND: 6 of 6 parts come from a retailer that does not state a delivery charge (SabrePC, Newegg, Newegg, ARCTIC, Newegg, Newegg), so the rig costs at least this and never less. Usable memory is below physical memory for at least one pool (OS/driver reservation; editable in Settings).	$113.66 (derived) DERIVEDCPU · EPYC 9015 · 8× Server Premier KSM64R52BD4-96MC 96GB DDR5-6400 RDIMM: cost per GB/sArithmetic on observed inputs only. No tunable assumption enters it. cost_per_gbs = landed_cost / memory_bandwidth_gbs = $34,917 / 307.2 GB/s = $113.66 per GB/s landed_cost 34916.84 USD — rig_landed = Σ part landed_total memory_bandwidth_gbs 307.2 GB/s — the published bandwidth of the rig's fastest memory pool Some part prices are older than 36 h. LOWER BOUND: 6 of 6 parts come from a retailer that does not state a delivery charge (SabrePC, Newegg, Newegg, ARCTIC, Newegg, Newegg), so the rig costs at least this and never less. Weights are read from the fastest pool first; a rig whose weights spill into a slower pool decodes below this figure.	$6,149 (estimate) ESTIMATECPU · EPYC 9015 · 8× Server Premier KSM64R52BD4-96MC 96GB DDR5-6400 RDIMM: cost per tok/sArithmetic that also depends on a tunable assumption, listed below. cost_per_tok_s = landed_cost / decode_tok_s = $34,917 / 5.68 tok/s = $6,148.99 per tok/s landed_cost 34916.84 USD — rig_landed = Σ part landed_total decode_tok_s 5.68 tok/s — GLM-5.3 decoding on this rig at the Plan's context Some part prices are older than 36 h. LOWER BOUND: 6 of 6 parts come from a retailer that does not state a delivery charge (SabrePC, Newegg, Newegg, ARCTIC, Newegg, Newegg), so the rig costs at least this and never less. MoE: only active parameters are read per token (SPEC §3). Effective bits include embeddings and higher-precision tensors in mixed quants. Assumes active weights are quantized like the file average. KV cache is re-read on every token (DECISIONS D-008). Batch size 1 (single user, interactive).
Priced from6 offers across 3 retailersoldeststale 44 h — past the 36 hour freshness windowopen the rig below for each part’s retailer, condition and route
21	2× RTX PRO 6000 Blackwell Max-Q Workstation Edition, 96GB · EPYC 9015 · 8× M321R4GA3BB6-CQK 32GB DDR5-4800 RDIMMDIY workstation · 7 parts · break down its parts	5 BLOCKERS	$29,773 (derived) DERIVED2× RTX PRO 6000 Blackwell Max-Q Workstation Edition, 96GB · EPYC 9015 · 8× M321R4GA3BB6-CQK 32GB DDR5-4800 RDIMM: landed costArithmetic on observed inputs only. No tunable assumption enters it. rig_landed = Σ part landed_total $638 + $895 + $1,828 + $25,442 + $75 + $192 + $704 = $29,773 AMD EPYC 9015 63809 ¢ — SabrePC · ship ASRock Rack GENOAD8QM3-2T/BCM 89458 ¢ — Newegg · ship Samsung M321R4GA3BB6-CQK 32GB DDR5-4800 RDIMM 182776 ¢ — SabrePC · ship NVIDIA RTX PRO 6000 Blackwell Max-Q Workstation Edition, 96GB 2544202 ¢ — SabrePC · ship ARCTIC Freezer 4U-SP5 7463 ¢ — ARCTIC · ship Phanteks Enthoo Pro 2 Server Edition 19191 ¢ — Newegg · ship Seasonic PRIME PX-2200 70371 ¢ — Newegg · ship Some part prices are older than 36 h. LOWER BOUND: 7 of 7 parts come from a retailer that does not state a delivery charge (SabrePC, Newegg, SabrePC, SabrePC, ARCTIC, Newegg, Newegg), so the rig costs at least this and never less.	Unavailable — Purchase + 5-year power needs the power term. Load draw at the wall is unavailable: Electrical feasibility cannot be computed exactly: no load or TDP watts for ASRock Rack GENOAD8QM3-2T/BCM, Samsung M321R4GA3BB6-CQK 32GB DDR5-4800 RDIMM, ARCTIC Freezer 4U-SP5, Phanteks Enthoo Pro 2 Server Edition. (missing: ASRock Rack GENOAD8QM3-2T/BCM: load_watts or tdp_watts, Samsung M321R4GA3BB6-CQK 32GB DDR5-4800 RDIMM: load_watts or tdp_watts, ARCTIC Freezer 4U-SP5: load_watts or tdp_watts, Phanteks Enthoo Pro 2 Server Edition: load_watts or tdp_watts, AMD EPYC 9015: idle_watts, ASRock Rack GENOAD8QM3-2T/BCM: idle_watts, Samsung M321R4GA3BB6-CQK 32GB DDR5-4800 RDIMM: idle_watts, NVIDIA RTX PRO 6000 Blackwell Max-Q Workstation Edition, 96GB: idle_watts, ARCTIC Freezer 4U-SP5: idle_watts, Phanteks Enthoo Pro 2 Server Edition: idle_watts, Seasonic PRIME PX-2200: idle_watts)	$69.33 (derived) DERIVED2× RTX PRO 6000 Blackwell Max-Q Workstation Edition, 96GB · EPYC 9015 · 8× M321R4GA3BB6-CQK 32GB DDR5-4800 RDIMM: cost per GBArithmetic on observed inputs only. No tunable assumption enters it. cost_per_gb = landed_cost / usable_memory_gb = $29,773 / 429.44 GB = $69.33 /GB landed_cost 29772.7 USD — rig_landed = Σ part landed_total usable_memory_gb 429.44 GB — usable memory after the OS and framework reservation (Settings → Estimator) Some part prices are older than 36 h. LOWER BOUND: 7 of 7 parts come from a retailer that does not state a delivery charge (SabrePC, Newegg, SabrePC, SabrePC, ARCTIC, Newegg, Newegg), so the rig costs at least this and never less. Usable memory is below physical memory for at least one pool (OS/driver reservation; editable in Settings).	$16.61 (derived) DERIVED2× RTX PRO 6000 Blackwell Max-Q Workstation Edition, 96GB · EPYC 9015 · 8× M321R4GA3BB6-CQK 32GB DDR5-4800 RDIMM: cost per GB/sArithmetic on observed inputs only. No tunable assumption enters it. cost_per_gbs = landed_cost / memory_bandwidth_gbs = $29,773 / 1,792 GB/s = $16.61 per GB/s landed_cost 29772.7 USD — rig_landed = Σ part landed_total memory_bandwidth_gbs 1792 GB/s — the published bandwidth of the rig's fastest memory pool Some part prices are older than 36 h. LOWER BOUND: 7 of 7 parts come from a retailer that does not state a delivery charge (SabrePC, Newegg, SabrePC, SabrePC, ARCTIC, Newegg, Newegg), so the rig costs at least this and never less. Weights are read from the fastest pool first; a rig whose weights spill into a slower pool decodes below this figure.	Unavailable — Cost per tok/s needs decode_tok_s: GLM-5.3 has no decode speed on this rig: Model needs 507.62 GB but DIY: 2× NVIDIA RTX PRO 6000 Blackwell Max-Q Workstation Edition, 96GB + AMD EPYC 9015 has 429.44 GB usable — it will not load, so there is no decode speed.
Priced from7 offers across 3 retailersoldeststale 44 h — past the 36 hour freshness windowopen the rig below for each part’s retailer, condition and route
22	Hybrid · RTX PRO 6000 Blackwell Workstation Edition, 96GB + EPYC 9015 · 8× M321R8GA0BB0-CQK 64GB DDR5-4800 RDIMMDIY workstation · 7 parts · break down its parts	4 BLOCKERS	$24,229 (derived) DERIVEDHybrid · RTX PRO 6000 Blackwell Workstation Edition, 96GB + EPYC 9015 · 8× M321R8GA0BB0-CQK 64GB DDR5-4800 RDIMM: landed costArithmetic on observed inputs only. No tunable assumption enters it. rig_landed = Σ part landed_total $638 + $895 + $2,534 + $19,192 + $75 + $192 + $704 = $24,229 AMD EPYC 9015 63809 ¢ — SabrePC · ship ASRock Rack GENOAD8QM3-2T/BCM 89458 ¢ — Newegg · ship Samsung M321R8GA0BB0-CQK 64GB DDR5-4800 RDIMM 253400 ¢ — SabrePC · ship NVIDIA RTX PRO 6000 Blackwell Workstation Edition, 96GB 1919249 ¢ — Newegg · ship ARCTIC Freezer 4U-SP5 7463 ¢ — ARCTIC · ship Phanteks Enthoo Pro 2 Server Edition 19191 ¢ — Newegg · ship Seasonic PRIME PX-2200 70371 ¢ — Newegg · ship Some part prices are older than 36 h. LOWER BOUND: 7 of 7 parts come from a retailer that does not state a delivery charge (SabrePC, Newegg, SabrePC, Newegg, ARCTIC, Newegg, Newegg), so the rig costs at least this and never less.	Unavailable — Purchase + 5-year power needs the power term. Load draw at the wall is unavailable: Electrical feasibility cannot be computed exactly: no load or TDP watts for ASRock Rack GENOAD8QM3-2T/BCM, Samsung M321R8GA0BB0-CQK 64GB DDR5-4800 RDIMM, ARCTIC Freezer 4U-SP5, Phanteks Enthoo Pro 2 Server Edition. (missing: ASRock Rack GENOAD8QM3-2T/BCM: load_watts or tdp_watts, Samsung M321R8GA0BB0-CQK 64GB DDR5-4800 RDIMM: load_watts or tdp_watts, ARCTIC Freezer 4U-SP5: load_watts or tdp_watts, Phanteks Enthoo Pro 2 Server Edition: load_watts or tdp_watts, AMD EPYC 9015: idle_watts, ASRock Rack GENOAD8QM3-2T/BCM: idle_watts, Samsung M321R8GA0BB0-CQK 64GB DDR5-4800 RDIMM: idle_watts, NVIDIA RTX PRO 6000 Blackwell Workstation Edition, 96GB: idle_watts, ARCTIC Freezer 4U-SP5: idle_watts, Phanteks Enthoo Pro 2 Server Edition: idle_watts, Seasonic PRIME PX-2200: idle_watts)	$41.81 (derived) DERIVEDHybrid · RTX PRO 6000 Blackwell Workstation Edition, 96GB + EPYC 9015 · 8× M321R8GA0BB0-CQK 64GB DDR5-4800 RDIMM: cost per GBArithmetic on observed inputs only. No tunable assumption enters it. cost_per_gb = landed_cost / usable_memory_gb = $24,229 / 579.52 GB = $41.81 /GB landed_cost 24229.41 USD — rig_landed = Σ part landed_total usable_memory_gb 579.52 GB — usable memory after the OS and framework reservation (Settings → Estimator) Some part prices are older than 36 h. LOWER BOUND: 7 of 7 parts come from a retailer that does not state a delivery charge (SabrePC, Newegg, SabrePC, Newegg, ARCTIC, Newegg, Newegg), so the rig costs at least this and never less. Usable memory is below physical memory for at least one pool (OS/driver reservation; editable in Settings).	$13.52 (derived) DERIVEDHybrid · RTX PRO 6000 Blackwell Workstation Edition, 96GB + EPYC 9015 · 8× M321R8GA0BB0-CQK 64GB DDR5-4800 RDIMM: cost per GB/sArithmetic on observed inputs only. No tunable assumption enters it. cost_per_gbs = landed_cost / memory_bandwidth_gbs = $24,229 / 1,792 GB/s = $13.52 per GB/s landed_cost 24229.41 USD — rig_landed = Σ part landed_total memory_bandwidth_gbs 1792 GB/s — the published bandwidth of the rig's fastest memory pool Some part prices are older than 36 h. LOWER BOUND: 7 of 7 parts come from a retailer that does not state a delivery charge (SabrePC, Newegg, SabrePC, Newegg, ARCTIC, Newegg, Newegg), so the rig costs at least this and never less. Weights are read from the fastest pool first; a rig whose weights spill into a slower pool decodes below this figure.	$3,637 (estimate) ESTIMATEHybrid · RTX PRO 6000 Blackwell Workstation Edition, 96GB + EPYC 9015 · 8× M321R8GA0BB0-CQK 64GB DDR5-4800 RDIMM: cost per tok/sArithmetic that also depends on a tunable assumption, listed below. cost_per_tok_s = landed_cost / decode_tok_s = $24,229 / 6.66 tok/s = $3,637.29 per tok/s landed_cost 24229.41 USD — rig_landed = Σ part landed_total decode_tok_s 6.66 tok/s — GLM-5.3 decoding on this rig at the Plan's context Some part prices are older than 36 h. LOWER BOUND: 7 of 7 parts come from a retailer that does not state a delivery charge (SabrePC, Newegg, SabrePC, Newegg, ARCTIC, Newegg, Newegg), so the rig costs at least this and never less. MoE: only active parameters are read per token (SPEC §3). Effective bits include embeddings and higher-precision tensors in mixed quants. Assumes active weights are quantized like the file average. KV cache is re-read on every token (DECISIONS D-008). Pools are read sequentially per token, so per-pool times add (harmonic combination). Per-token overhead on a split placement is DERIVED, not measured: each pool's own single-pool figure is charged in proportion to the share of each token it serves, because the fixed cost is per operation and the split puts that share of the operations on each device. Calibrated with that pool serving the whole model (D-013): GPU VRAM (1.58 ms), System RAM (44.61 ms). Batch size 1 (single user, interactive).
Priced from7 offers across 3 retailersoldeststale 44 h — past the 36 hour freshness windowopen the rig below for each part’s retailer, condition and route
23	1× RTX PRO 6000 Blackwell Workstation Edition, 96GB · EPYC 9015 · 8× M321R4GA3BB6-CQK 32GB DDR5-4800 RDIMMDIY workstation · 7 parts · break down its parts	5 BLOCKERS	$23,523 (derived) DERIVED1× RTX PRO 6000 Blackwell Workstation Edition, 96GB · EPYC 9015 · 8× M321R4GA3BB6-CQK 32GB DDR5-4800 RDIMM: landed costArithmetic on observed inputs only. No tunable assumption enters it. rig_landed = Σ part landed_total $638 + $895 + $1,828 + $19,192 + $75 + $192 + $704 = $23,523 AMD EPYC 9015 63809 ¢ — SabrePC · ship ASRock Rack GENOAD8QM3-2T/BCM 89458 ¢ — Newegg · ship Samsung M321R4GA3BB6-CQK 32GB DDR5-4800 RDIMM 182776 ¢ — SabrePC · ship NVIDIA RTX PRO 6000 Blackwell Workstation Edition, 96GB 1919249 ¢ — Newegg · ship ARCTIC Freezer 4U-SP5 7463 ¢ — ARCTIC · ship Phanteks Enthoo Pro 2 Server Edition 19191 ¢ — Newegg · ship Seasonic PRIME PX-2200 70371 ¢ — Newegg · ship Some part prices are older than 36 h. LOWER BOUND: 7 of 7 parts come from a retailer that does not state a delivery charge (SabrePC, Newegg, SabrePC, Newegg, ARCTIC, Newegg, Newegg), so the rig costs at least this and never less.	Unavailable — Purchase + 5-year power needs the power term. Load draw at the wall is unavailable: Electrical feasibility cannot be computed exactly: no load or TDP watts for ASRock Rack GENOAD8QM3-2T/BCM, Samsung M321R4GA3BB6-CQK 32GB DDR5-4800 RDIMM, ARCTIC Freezer 4U-SP5, Phanteks Enthoo Pro 2 Server Edition. (missing: ASRock Rack GENOAD8QM3-2T/BCM: load_watts or tdp_watts, Samsung M321R4GA3BB6-CQK 32GB DDR5-4800 RDIMM: load_watts or tdp_watts, ARCTIC Freezer 4U-SP5: load_watts or tdp_watts, Phanteks Enthoo Pro 2 Server Edition: load_watts or tdp_watts, AMD EPYC 9015: idle_watts, ASRock Rack GENOAD8QM3-2T/BCM: idle_watts, Samsung M321R4GA3BB6-CQK 32GB DDR5-4800 RDIMM: idle_watts, NVIDIA RTX PRO 6000 Blackwell Workstation Edition, 96GB: idle_watts, ARCTIC Freezer 4U-SP5: idle_watts, Phanteks Enthoo Pro 2 Server Edition: idle_watts, Seasonic PRIME PX-2200: idle_watts)	$69.94 (derived) DERIVED1× RTX PRO 6000 Blackwell Workstation Edition, 96GB · EPYC 9015 · 8× M321R4GA3BB6-CQK 32GB DDR5-4800 RDIMM: cost per GBArithmetic on observed inputs only. No tunable assumption enters it. cost_per_gb = landed_cost / usable_memory_gb = $23,523 / 336.32 GB = $69.94 /GB landed_cost 23523.17 USD — rig_landed = Σ part landed_total usable_memory_gb 336.32 GB — usable memory after the OS and framework reservation (Settings → Estimator) Some part prices are older than 36 h. LOWER BOUND: 7 of 7 parts come from a retailer that does not state a delivery charge (SabrePC, Newegg, SabrePC, Newegg, ARCTIC, Newegg, Newegg), so the rig costs at least this and never less. Usable memory is below physical memory for at least one pool (OS/driver reservation; editable in Settings).	$13.13 (derived) DERIVED1× RTX PRO 6000 Blackwell Workstation Edition, 96GB · EPYC 9015 · 8× M321R4GA3BB6-CQK 32GB DDR5-4800 RDIMM: cost per GB/sArithmetic on observed inputs only. No tunable assumption enters it. cost_per_gbs = landed_cost / memory_bandwidth_gbs = $23,523 / 1,792 GB/s = $13.13 per GB/s landed_cost 23523.17 USD — rig_landed = Σ part landed_total memory_bandwidth_gbs 1792 GB/s — the published bandwidth of the rig's fastest memory pool Some part prices are older than 36 h. LOWER BOUND: 7 of 7 parts come from a retailer that does not state a delivery charge (SabrePC, Newegg, SabrePC, Newegg, ARCTIC, Newegg, Newegg), so the rig costs at least this and never less. Weights are read from the fastest pool first; a rig whose weights spill into a slower pool decodes below this figure.	Unavailable — Cost per tok/s needs decode_tok_s: GLM-5.3 has no decode speed on this rig: Model needs 507.62 GB but DIY: NVIDIA RTX PRO 6000 Blackwell Workstation Edition, 96GB + AMD EPYC 9015 has 336.32 GB usable — it will not load, so there is no decode speed.
Priced from7 offers across 3 retailersoldeststale 44 h — past the 36 hour freshness windowopen the rig below for each part’s retailer, condition and route
24	Hybrid · RTX PRO 6000 Blackwell Max-Q Workstation Edition, 96GB + EPYC 9015 · 8× M321R8GA0BB0-CQK 64GB DDR5-4800 RDIMMDIY workstation · 7 parts · break down its parts	4 BLOCKERS	$17,758 (derived) DERIVEDHybrid · RTX PRO 6000 Blackwell Max-Q Workstation Edition, 96GB + EPYC 9015 · 8× M321R8GA0BB0-CQK 64GB DDR5-4800 RDIMM: landed costArithmetic on observed inputs only. No tunable assumption enters it. rig_landed = Σ part landed_total $638 + $895 + $2,534 + $12,721 + $75 + $192 + $704 = $17,758 AMD EPYC 9015 63809 ¢ — SabrePC · ship ASRock Rack GENOAD8QM3-2T/BCM 89458 ¢ — Newegg · ship Samsung M321R8GA0BB0-CQK 64GB DDR5-4800 RDIMM 253400 ¢ — SabrePC · ship NVIDIA RTX PRO 6000 Blackwell Max-Q Workstation Edition, 96GB 1272101 ¢ — SabrePC · ship ARCTIC Freezer 4U-SP5 7463 ¢ — ARCTIC · ship Phanteks Enthoo Pro 2 Server Edition 19191 ¢ — Newegg · ship Seasonic PRIME PX-2200 70371 ¢ — Newegg · ship Some part prices are older than 36 h. LOWER BOUND: 7 of 7 parts come from a retailer that does not state a delivery charge (SabrePC, Newegg, SabrePC, SabrePC, ARCTIC, Newegg, Newegg), so the rig costs at least this and never less.	Unavailable — Purchase + 5-year power needs the power term. Load draw at the wall is unavailable: Electrical feasibility cannot be computed exactly: no load or TDP watts for ASRock Rack GENOAD8QM3-2T/BCM, Samsung M321R8GA0BB0-CQK 64GB DDR5-4800 RDIMM, ARCTIC Freezer 4U-SP5, Phanteks Enthoo Pro 2 Server Edition. (missing: ASRock Rack GENOAD8QM3-2T/BCM: load_watts or tdp_watts, Samsung M321R8GA0BB0-CQK 64GB DDR5-4800 RDIMM: load_watts or tdp_watts, ARCTIC Freezer 4U-SP5: load_watts or tdp_watts, Phanteks Enthoo Pro 2 Server Edition: load_watts or tdp_watts, AMD EPYC 9015: idle_watts, ASRock Rack GENOAD8QM3-2T/BCM: idle_watts, Samsung M321R8GA0BB0-CQK 64GB DDR5-4800 RDIMM: idle_watts, NVIDIA RTX PRO 6000 Blackwell Max-Q Workstation Edition, 96GB: idle_watts, ARCTIC Freezer 4U-SP5: idle_watts, Phanteks Enthoo Pro 2 Server Edition: idle_watts, Seasonic PRIME PX-2200: idle_watts)	$30.64 (derived) DERIVEDHybrid · RTX PRO 6000 Blackwell Max-Q Workstation Edition, 96GB + EPYC 9015 · 8× M321R8GA0BB0-CQK 64GB DDR5-4800 RDIMM: cost per GBArithmetic on observed inputs only. No tunable assumption enters it. cost_per_gb = landed_cost / usable_memory_gb = $17,758 / 579.52 GB = $30.64 /GB landed_cost 17757.93 USD — rig_landed = Σ part landed_total usable_memory_gb 579.52 GB — usable memory after the OS and framework reservation (Settings → Estimator) Some part prices are older than 36 h. LOWER BOUND: 7 of 7 parts come from a retailer that does not state a delivery charge (SabrePC, Newegg, SabrePC, SabrePC, ARCTIC, Newegg, Newegg), so the rig costs at least this and never less. Usable memory is below physical memory for at least one pool (OS/driver reservation; editable in Settings).	$9.91 (derived) DERIVEDHybrid · RTX PRO 6000 Blackwell Max-Q Workstation Edition, 96GB + EPYC 9015 · 8× M321R8GA0BB0-CQK 64GB DDR5-4800 RDIMM: cost per GB/sArithmetic on observed inputs only. No tunable assumption enters it. cost_per_gbs = landed_cost / memory_bandwidth_gbs = $17,758 / 1,792 GB/s = $9.91 per GB/s landed_cost 17757.93 USD — rig_landed = Σ part landed_total memory_bandwidth_gbs 1792 GB/s — the published bandwidth of the rig's fastest memory pool Some part prices are older than 36 h. LOWER BOUND: 7 of 7 parts come from a retailer that does not state a delivery charge (SabrePC, Newegg, SabrePC, SabrePC, ARCTIC, Newegg, Newegg), so the rig costs at least this and never less. Weights are read from the fastest pool first; a rig whose weights spill into a slower pool decodes below this figure.	$2,666 (estimate) ESTIMATEHybrid · RTX PRO 6000 Blackwell Max-Q Workstation Edition, 96GB + EPYC 9015 · 8× M321R8GA0BB0-CQK 64GB DDR5-4800 RDIMM: cost per tok/sArithmetic that also depends on a tunable assumption, listed below. cost_per_tok_s = landed_cost / decode_tok_s = $17,758 / 6.66 tok/s = $2,665.8 per tok/s landed_cost 17757.93 USD — rig_landed = Σ part landed_total decode_tok_s 6.66 tok/s — GLM-5.3 decoding on this rig at the Plan's context Some part prices are older than 36 h. LOWER BOUND: 7 of 7 parts come from a retailer that does not state a delivery charge (SabrePC, Newegg, SabrePC, SabrePC, ARCTIC, Newegg, Newegg), so the rig costs at least this and never less. MoE: only active parameters are read per token (SPEC §3). Effective bits include embeddings and higher-precision tensors in mixed quants. Assumes active weights are quantized like the file average. KV cache is re-read on every token (DECISIONS D-008). Pools are read sequentially per token, so per-pool times add (harmonic combination). Per-token overhead on a split placement is DERIVED, not measured: each pool's own single-pool figure is charged in proportion to the share of each token it serves, because the fixed cost is per operation and the split puts that share of the operations on each device. Calibrated with that pool serving the whole model (D-013): GPU VRAM (1.58 ms), System RAM (44.61 ms). Batch size 1 (single user, interactive).
Priced from7 offers across 3 retailersoldeststale 44 h — past the 36 hour freshness windowopen the rig below for each part’s retailer, condition and route
25	1× RTX PRO 6000 Blackwell Max-Q Workstation Edition, 96GB · EPYC 9015 · 8× M321R4GA3BB6-CQK 32GB DDR5-4800 RDIMMDIY workstation · 7 parts · break down its parts	5 BLOCKERS	$17,052 (derived) DERIVED1× RTX PRO 6000 Blackwell Max-Q Workstation Edition, 96GB · EPYC 9015 · 8× M321R4GA3BB6-CQK 32GB DDR5-4800 RDIMM: landed costArithmetic on observed inputs only. No tunable assumption enters it. rig_landed = Σ part landed_total $638 + $895 + $1,828 + $12,721 + $75 + $192 + $704 = $17,052 AMD EPYC 9015 63809 ¢ — SabrePC · ship ASRock Rack GENOAD8QM3-2T/BCM 89458 ¢ — Newegg · ship Samsung M321R4GA3BB6-CQK 32GB DDR5-4800 RDIMM 182776 ¢ — SabrePC · ship NVIDIA RTX PRO 6000 Blackwell Max-Q Workstation Edition, 96GB 1272101 ¢ — SabrePC · ship ARCTIC Freezer 4U-SP5 7463 ¢ — ARCTIC · ship Phanteks Enthoo Pro 2 Server Edition 19191 ¢ — Newegg · ship Seasonic PRIME PX-2200 70371 ¢ — Newegg · ship Some part prices are older than 36 h. LOWER BOUND: 7 of 7 parts come from a retailer that does not state a delivery charge (SabrePC, Newegg, SabrePC, SabrePC, ARCTIC, Newegg, Newegg), so the rig costs at least this and never less.	Unavailable — Purchase + 5-year power needs the power term. Load draw at the wall is unavailable: Electrical feasibility cannot be computed exactly: no load or TDP watts for ASRock Rack GENOAD8QM3-2T/BCM, Samsung M321R4GA3BB6-CQK 32GB DDR5-4800 RDIMM, ARCTIC Freezer 4U-SP5, Phanteks Enthoo Pro 2 Server Edition. (missing: ASRock Rack GENOAD8QM3-2T/BCM: load_watts or tdp_watts, Samsung M321R4GA3BB6-CQK 32GB DDR5-4800 RDIMM: load_watts or tdp_watts, ARCTIC Freezer 4U-SP5: load_watts or tdp_watts, Phanteks Enthoo Pro 2 Server Edition: load_watts or tdp_watts, AMD EPYC 9015: idle_watts, ASRock Rack GENOAD8QM3-2T/BCM: idle_watts, Samsung M321R4GA3BB6-CQK 32GB DDR5-4800 RDIMM: idle_watts, NVIDIA RTX PRO 6000 Blackwell Max-Q Workstation Edition, 96GB: idle_watts, ARCTIC Freezer 4U-SP5: idle_watts, Phanteks Enthoo Pro 2 Server Edition: idle_watts, Seasonic PRIME PX-2200: idle_watts)	$50.70 (derived) DERIVED1× RTX PRO 6000 Blackwell Max-Q Workstation Edition, 96GB · EPYC 9015 · 8× M321R4GA3BB6-CQK 32GB DDR5-4800 RDIMM: cost per GBArithmetic on observed inputs only. No tunable assumption enters it. cost_per_gb = landed_cost / usable_memory_gb = $17,052 / 336.32 GB = $50.7 /GB landed_cost 17051.69 USD — rig_landed = Σ part landed_total usable_memory_gb 336.32 GB — usable memory after the OS and framework reservation (Settings → Estimator) Some part prices are older than 36 h. LOWER BOUND: 7 of 7 parts come from a retailer that does not state a delivery charge (SabrePC, Newegg, SabrePC, SabrePC, ARCTIC, Newegg, Newegg), so the rig costs at least this and never less. Usable memory is below physical memory for at least one pool (OS/driver reservation; editable in Settings).	$9.52 (derived) DERIVED1× RTX PRO 6000 Blackwell Max-Q Workstation Edition, 96GB · EPYC 9015 · 8× M321R4GA3BB6-CQK 32GB DDR5-4800 RDIMM: cost per GB/sArithmetic on observed inputs only. No tunable assumption enters it. cost_per_gbs = landed_cost / memory_bandwidth_gbs = $17,052 / 1,792 GB/s = $9.52 per GB/s landed_cost 17051.69 USD — rig_landed = Σ part landed_total memory_bandwidth_gbs 1792 GB/s — the published bandwidth of the rig's fastest memory pool Some part prices are older than 36 h. LOWER BOUND: 7 of 7 parts come from a retailer that does not state a delivery charge (SabrePC, Newegg, SabrePC, SabrePC, ARCTIC, Newegg, Newegg), so the rig costs at least this and never less. Weights are read from the fastest pool first; a rig whose weights spill into a slower pool decodes below this figure.	Unavailable — Cost per tok/s needs decode_tok_s: GLM-5.3 has no decode speed on this rig: Model needs 507.62 GB but DIY: NVIDIA RTX PRO 6000 Blackwell Max-Q Workstation Edition, 96GB + AMD EPYC 9015 has 336.32 GB usable — it will not load, so there is no decode speed.
Priced from7 offers across 3 retailersoldeststale 44 h — past the 36 hour freshness windowopen the rig below for each part’s retailer, condition and route
26	2× GeForce RTX 5090 32G GAMING TRIO OC · EPYC 9015 · 8× M321R4GA3BB6-CQK 32GB DDR5-4800 RDIMMDIY workstation · 7 parts · break down its parts	5 BLOCKERS	$15,206 (derived) DERIVED2× GeForce RTX 5090 32G GAMING TRIO OC · EPYC 9015 · 8× M321R4GA3BB6-CQK 32GB DDR5-4800 RDIMM: landed costArithmetic on observed inputs only. No tunable assumption enters it. rig_landed = Σ part landed_total $638 + $895 + $1,828 + $10,876 + $75 + $192 + $704 = $15,206 AMD EPYC 9015 63809 ¢ — SabrePC · ship ASRock Rack GENOAD8QM3-2T/BCM 89458 ¢ — Newegg · ship Samsung M321R4GA3BB6-CQK 32GB DDR5-4800 RDIMM 182776 ¢ — SabrePC · ship MSI GeForce RTX 5090 32G GAMING TRIO OC 1087572 ¢ — Newegg · ship ARCTIC Freezer 4U-SP5 7463 ¢ — ARCTIC · ship Phanteks Enthoo Pro 2 Server Edition 19191 ¢ — Newegg · ship Seasonic PRIME PX-2200 70371 ¢ — Newegg · ship Some part prices are older than 36 h. LOWER BOUND: 7 of 7 parts come from a retailer that does not state a delivery charge (SabrePC, Newegg, SabrePC, Newegg, ARCTIC, Newegg, Newegg), so the rig costs at least this and never less.	Unavailable — Purchase + 5-year power needs the power term. Load draw at the wall is unavailable: Electrical feasibility cannot be computed exactly: no load or TDP watts for ASRock Rack GENOAD8QM3-2T/BCM, Samsung M321R4GA3BB6-CQK 32GB DDR5-4800 RDIMM, ARCTIC Freezer 4U-SP5, Phanteks Enthoo Pro 2 Server Edition. (missing: ASRock Rack GENOAD8QM3-2T/BCM: load_watts or tdp_watts, Samsung M321R4GA3BB6-CQK 32GB DDR5-4800 RDIMM: load_watts or tdp_watts, ARCTIC Freezer 4U-SP5: load_watts or tdp_watts, Phanteks Enthoo Pro 2 Server Edition: load_watts or tdp_watts, AMD EPYC 9015: idle_watts, ASRock Rack GENOAD8QM3-2T/BCM: idle_watts, Samsung M321R4GA3BB6-CQK 32GB DDR5-4800 RDIMM: idle_watts, MSI GeForce RTX 5090 32G GAMING TRIO OC: idle_watts, ARCTIC Freezer 4U-SP5: idle_watts, Phanteks Enthoo Pro 2 Server Edition: idle_watts, Seasonic PRIME PX-2200: idle_watts)	$49.81 (derived) DERIVED2× GeForce RTX 5090 32G GAMING TRIO OC · EPYC 9015 · 8× M321R4GA3BB6-CQK 32GB DDR5-4800 RDIMM: cost per GBArithmetic on observed inputs only. No tunable assumption enters it. cost_per_gb = landed_cost / usable_memory_gb = $15,206 / 305.28 GB = $49.81 /GB landed_cost 15206.4 USD — rig_landed = Σ part landed_total usable_memory_gb 305.28 GB — usable memory after the OS and framework reservation (Settings → Estimator) Some part prices are older than 36 h. LOWER BOUND: 7 of 7 parts come from a retailer that does not state a delivery charge (SabrePC, Newegg, SabrePC, Newegg, ARCTIC, Newegg, Newegg), so the rig costs at least this and never less. Usable memory is below physical memory for at least one pool (OS/driver reservation; editable in Settings).	$8.49 (derived) DERIVED2× GeForce RTX 5090 32G GAMING TRIO OC · EPYC 9015 · 8× M321R4GA3BB6-CQK 32GB DDR5-4800 RDIMM: cost per GB/sArithmetic on observed inputs only. No tunable assumption enters it. cost_per_gbs = landed_cost / memory_bandwidth_gbs = $15,206 / 1,792 GB/s = $8.49 per GB/s landed_cost 15206.4 USD — rig_landed = Σ part landed_total memory_bandwidth_gbs 1792 GB/s — the published bandwidth of the rig's fastest memory pool Some part prices are older than 36 h. LOWER BOUND: 7 of 7 parts come from a retailer that does not state a delivery charge (SabrePC, Newegg, SabrePC, Newegg, ARCTIC, Newegg, Newegg), so the rig costs at least this and never less. Weights are read from the fastest pool first; a rig whose weights spill into a slower pool decodes below this figure.	Unavailable — Cost per tok/s needs decode_tok_s: GLM-5.3 has no decode speed on this rig: Model needs 507.62 GB but DIY: 2× MSI GeForce RTX 5090 32G GAMING TRIO OC + AMD EPYC 9015 has 305.28 GB usable — it will not load, so there is no decode speed.
Priced from7 offers across 3 retailersoldeststale 44 h — past the 36 hour freshness windowopen the rig below for each part’s retailer, condition and route
27	Apple Mac Studio (2026), M5 Ultra 36-core CPU / 80-core GPU, 256GBsealed appliance · 1 part · break down its parts	5 BLOCKERS	$11,514 (derived) DERIVEDApple Mac Studio (2026), M5 Ultra 36-core CPU / 80-core GPU, 256GB: landed costArithmetic on observed inputs only. No tunable assumption enters it. rig_landed = Σ part landed_total $11,514 = $11,514 Apple Mac Studio (2026), M5 Ultra 36-core CPU / 80-core GPU, 256GB 1151443 ¢ — Apple · ship LOWER BOUND: 1 of 1 parts come from a retailer that does not state a delivery charge (Apple), so the rig costs at least this and never less.	$12,709 (estimate) ESTIMATEApple Mac Studio (2026), M5 Ultra 36-core CPU / 80-core GPU, 256GB: purchase plus five years of powerArithmetic that also depends on a tunable assumption, listed below. purchase_plus_5yr_power = purchase + five_year_power purchase_plus_5yr_power = $11,514.43 + $1,194.47 = $12,708.90 purchase_cents 1151443 ¢ — rig_landed = Σ part landed_total five_year_power_cents 119447 ¢ — five_year_power = annual_kwh × ¢_per_kwh × 5 LOWER BOUND: 1 of 1 parts come from a retailer that does not state a delivery charge (Apple), so the rig costs at least this and never less. Watts are at the wall (PSU conversion losses included by the caller). Load and idle draws are averages over their hours, not peaks. Duty cycle (Settings → Power): 6 h/day at load, 18 h/day idle. 365 days per year (leap days ignored, −0.07%). Flat rate for the whole period: no escalation, no time-of-use or tiered pricing, no demand charges. Rounded to the nearest cent, half-up, once from the unrounded annual cost.	$59.97 (derived) DERIVEDApple Mac Studio (2026), M5 Ultra 36-core CPU / 80-core GPU, 256GB: cost per GBArithmetic on observed inputs only. No tunable assumption enters it. cost_per_gb = landed_cost / usable_memory_gb = $11,514 / 192 GB = $59.97 /GB landed_cost 11514.43 USD — rig_landed = Σ part landed_total usable_memory_gb 192 GB — usable memory after the OS and framework reservation (Settings → Estimator) LOWER BOUND: 1 of 1 parts come from a retailer that does not state a delivery charge (Apple), so the rig costs at least this and never less. Usable memory is below physical memory for at least one pool (OS/driver reservation; editable in Settings).	$9.60 (derived) DERIVEDApple Mac Studio (2026), M5 Ultra 36-core CPU / 80-core GPU, 256GB: cost per GB/sArithmetic on observed inputs only. No tunable assumption enters it. cost_per_gbs = landed_cost / memory_bandwidth_gbs = $11,514 / 1,200 GB/s = $9.6 per GB/s landed_cost 11514.43 USD — rig_landed = Σ part landed_total memory_bandwidth_gbs 1200 GB/s — the published bandwidth of the rig's fastest memory pool LOWER BOUND: 1 of 1 parts come from a retailer that does not state a delivery charge (Apple), so the rig costs at least this and never less. Weights are read from the fastest pool first; a rig whose weights spill into a slower pool decodes below this figure.	Unavailable — Cost per tok/s needs decode_tok_s: GLM-5.3 has no decode speed on this rig: Model needs 507.62 GB but Apple Mac Studio (2026), M5 Ultra 36-core CPU / 80-core GPU, 256GB has 192 GB usable — it will not load, so there is no decode speed.
Priced fromApplecondition newaccess tier 3, route apple_store30 hshipped
28	Hybrid · GeForce RTX 5090 32G GAMING TRIO OC + EPYC 9015 · 8× M321R8GA0BB0-CQK 64GB DDR5-4800 RDIMMDIY workstation · 7 parts · break down its parts	4 BLOCKERS	$10,475 (derived) DERIVEDHybrid · GeForce RTX 5090 32G GAMING TRIO OC + EPYC 9015 · 8× M321R8GA0BB0-CQK 64GB DDR5-4800 RDIMM: landed costArithmetic on observed inputs only. No tunable assumption enters it. rig_landed = Σ part landed_total $638 + $895 + $2,534 + $5,438 + $75 + $192 + $704 = $10,475 AMD EPYC 9015 63809 ¢ — SabrePC · ship ASRock Rack GENOAD8QM3-2T/BCM 89458 ¢ — Newegg · ship Samsung M321R8GA0BB0-CQK 64GB DDR5-4800 RDIMM 253400 ¢ — SabrePC · ship MSI GeForce RTX 5090 32G GAMING TRIO OC 543786 ¢ — Newegg · ship ARCTIC Freezer 4U-SP5 7463 ¢ — ARCTIC · ship Phanteks Enthoo Pro 2 Server Edition 19191 ¢ — Newegg · ship Seasonic PRIME PX-2200 70371 ¢ — Newegg · ship Some part prices are older than 36 h. LOWER BOUND: 7 of 7 parts come from a retailer that does not state a delivery charge (SabrePC, Newegg, SabrePC, Newegg, ARCTIC, Newegg, Newegg), so the rig costs at least this and never less.	Unavailable — Purchase + 5-year power needs the power term. Load draw at the wall is unavailable: Electrical feasibility cannot be computed exactly: no load or TDP watts for ASRock Rack GENOAD8QM3-2T/BCM, Samsung M321R8GA0BB0-CQK 64GB DDR5-4800 RDIMM, ARCTIC Freezer 4U-SP5, Phanteks Enthoo Pro 2 Server Edition. (missing: ASRock Rack GENOAD8QM3-2T/BCM: load_watts or tdp_watts, Samsung M321R8GA0BB0-CQK 64GB DDR5-4800 RDIMM: load_watts or tdp_watts, ARCTIC Freezer 4U-SP5: load_watts or tdp_watts, Phanteks Enthoo Pro 2 Server Edition: load_watts or tdp_watts, AMD EPYC 9015: idle_watts, ASRock Rack GENOAD8QM3-2T/BCM: idle_watts, Samsung M321R8GA0BB0-CQK 64GB DDR5-4800 RDIMM: idle_watts, MSI GeForce RTX 5090 32G GAMING TRIO OC: idle_watts, ARCTIC Freezer 4U-SP5: idle_watts, Phanteks Enthoo Pro 2 Server Edition: idle_watts, Seasonic PRIME PX-2200: idle_watts)	$20.24 (derived) DERIVEDHybrid · GeForce RTX 5090 32G GAMING TRIO OC + EPYC 9015 · 8× M321R8GA0BB0-CQK 64GB DDR5-4800 RDIMM: cost per GBArithmetic on observed inputs only. No tunable assumption enters it. cost_per_gb = landed_cost / usable_memory_gb = $10,475 / 517.44 GB = $20.24 /GB landed_cost 10474.78 USD — rig_landed = Σ part landed_total usable_memory_gb 517.44 GB — usable memory after the OS and framework reservation (Settings → Estimator) Some part prices are older than 36 h. LOWER BOUND: 7 of 7 parts come from a retailer that does not state a delivery charge (SabrePC, Newegg, SabrePC, Newegg, ARCTIC, Newegg, Newegg), so the rig costs at least this and never less. Usable memory is below physical memory for at least one pool (OS/driver reservation; editable in Settings).	$5.85 (derived) DERIVEDHybrid · GeForce RTX 5090 32G GAMING TRIO OC + EPYC 9015 · 8× M321R8GA0BB0-CQK 64GB DDR5-4800 RDIMM: cost per GB/sArithmetic on observed inputs only. No tunable assumption enters it. cost_per_gbs = landed_cost / memory_bandwidth_gbs = $10,475 / 1,792 GB/s = $5.85 per GB/s landed_cost 10474.78 USD — rig_landed = Σ part landed_total memory_bandwidth_gbs 1792 GB/s — the published bandwidth of the rig's fastest memory pool Some part prices are older than 36 h. LOWER BOUND: 7 of 7 parts come from a retailer that does not state a delivery charge (SabrePC, Newegg, SabrePC, Newegg, ARCTIC, Newegg, Newegg), so the rig costs at least this and never less. Weights are read from the fastest pool first; a rig whose weights spill into a slower pool decodes below this figure.	$1,754 (estimate) ESTIMATEHybrid · GeForce RTX 5090 32G GAMING TRIO OC + EPYC 9015 · 8× M321R8GA0BB0-CQK 64GB DDR5-4800 RDIMM: cost per tok/sArithmetic that also depends on a tunable assumption, listed below. cost_per_tok_s = landed_cost / decode_tok_s = $10,475 / 5.97 tok/s = $1,753.92 per tok/s landed_cost 10474.78 USD — rig_landed = Σ part landed_total decode_tok_s 5.97 tok/s — GLM-5.3 decoding on this rig at the Plan's context Some part prices are older than 36 h. LOWER BOUND: 7 of 7 parts come from a retailer that does not state a delivery charge (SabrePC, Newegg, SabrePC, Newegg, ARCTIC, Newegg, Newegg), so the rig costs at least this and never less. MoE: only active parameters are read per token (SPEC §3). Effective bits include embeddings and higher-precision tensors in mixed quants. Assumes active weights are quantized like the file average. KV cache is re-read on every token (DECISIONS D-008). Pools are read sequentially per token, so per-pool times add (harmonic combination). Per-token overhead on a split placement is DERIVED, not measured: each pool's own single-pool figure is charged in proportion to the share of each token it serves, because the fixed cost is per operation and the split puts that share of the operations on each device. Calibrated with that pool serving the whole model (D-013): GPU VRAM (1.58 ms), System RAM (44.61 ms). Batch size 1 (single user, interactive).
Priced from7 offers across 3 retailersoldeststale 44 h — past the 36 hour freshness windowopen the rig below for each part’s retailer, condition and route
29	Apple Mac Studio (2026), M5 Ultra 30-core CPU / 64-core GPU, 256GBsealed appliance · 1 part · break down its parts	5 BLOCKERS	$10,128 (derived) DERIVEDApple Mac Studio (2026), M5 Ultra 30-core CPU / 64-core GPU, 256GB: landed costArithmetic on observed inputs only. No tunable assumption enters it. rig_landed = Σ part landed_total $10,128 = $10,128 Apple Mac Studio (2026), M5 Ultra 30-core CPU / 64-core GPU, 256GB 1012831 ¢ — Apple · ship LOWER BOUND: 1 of 1 parts come from a retailer that does not state a delivery charge (Apple), so the rig costs at least this and never less.	Unavailable — Purchase + 5-year power needs the power term. Idle draw is not published for 1 of 1 part in this rig, and the duty cycle in Settings spends most of the day idle — counting an unknown idle draw as zero would understate the bill. (missing: Apple Mac Studio (2026), M5 Ultra 30-core CPU / 64-core GPU, 256GB: idle_watts)	$52.75 (derived) DERIVEDApple Mac Studio (2026), M5 Ultra 30-core CPU / 64-core GPU, 256GB: cost per GBArithmetic on observed inputs only. No tunable assumption enters it. cost_per_gb = landed_cost / usable_memory_gb = $10,128 / 192 GB = $52.75 /GB landed_cost 10128.31 USD — rig_landed = Σ part landed_total usable_memory_gb 192 GB — usable memory after the OS and framework reservation (Settings → Estimator) LOWER BOUND: 1 of 1 parts come from a retailer that does not state a delivery charge (Apple), so the rig costs at least this and never less. Usable memory is below physical memory for at least one pool (OS/driver reservation; editable in Settings).	$8.44 (derived) DERIVEDApple Mac Studio (2026), M5 Ultra 30-core CPU / 64-core GPU, 256GB: cost per GB/sArithmetic on observed inputs only. No tunable assumption enters it. cost_per_gbs = landed_cost / memory_bandwidth_gbs = $10,128 / 1,200 GB/s = $8.44 per GB/s landed_cost 10128.31 USD — rig_landed = Σ part landed_total memory_bandwidth_gbs 1200 GB/s — the published bandwidth of the rig's fastest memory pool LOWER BOUND: 1 of 1 parts come from a retailer that does not state a delivery charge (Apple), so the rig costs at least this and never less. Weights are read from the fastest pool first; a rig whose weights spill into a slower pool decodes below this figure.	Unavailable — Cost per tok/s needs decode_tok_s: GLM-5.3 has no decode speed on this rig: Model needs 507.62 GB but Apple Mac Studio (2026), M5 Ultra 30-core CPU / 64-core GPU, 256GB has 192 GB usable — it will not load, so there is no decode speed.
Priced fromApplecondition newaccess tier 3, route apple_store30 hshipped
30	3× Radeon AI PRO R9700 32GB (32358-01-20G) · EPYC 9015 · 8× M321R4GA3BB6-CQK 32GB DDR5-4800 RDIMMDIY workstation · 7 parts · break down its parts	6 BLOCKERS	$10,088 (derived) DERIVED3× Radeon AI PRO R9700 32GB (32358-01-20G) · EPYC 9015 · 8× M321R4GA3BB6-CQK 32GB DDR5-4800 RDIMM: landed costArithmetic on observed inputs only. No tunable assumption enters it. rig_landed = Σ part landed_total $638 + $895 + $1,828 + $5,758 + $75 + $192 + $704 = $10,088 AMD EPYC 9015 63809 ¢ — SabrePC · ship ASRock Rack GENOAD8QM3-2T/BCM 89458 ¢ — Newegg · ship Samsung M321R4GA3BB6-CQK 32GB DDR5-4800 RDIMM 182776 ¢ — SabrePC · ship Sapphire Radeon AI PRO R9700 32GB (32358-01-20G) 575772 ¢ — Newegg · ship ARCTIC Freezer 4U-SP5 7463 ¢ — ARCTIC · ship Phanteks Enthoo Pro 2 Server Edition 19191 ¢ — Newegg · ship Seasonic PRIME PX-2200 70371 ¢ — Newegg · ship Some part prices are older than 36 h. LOWER BOUND: 7 of 7 parts come from a retailer that does not state a delivery charge (SabrePC, Newegg, SabrePC, Newegg, ARCTIC, Newegg, Newegg), so the rig costs at least this and never less.	Unavailable — Purchase + 5-year power needs the power term. Load draw at the wall is unavailable: Electrical feasibility cannot be computed exactly: no load or TDP watts for ASRock Rack GENOAD8QM3-2T/BCM, Samsung M321R4GA3BB6-CQK 32GB DDR5-4800 RDIMM, ARCTIC Freezer 4U-SP5, Phanteks Enthoo Pro 2 Server Edition. (missing: ASRock Rack GENOAD8QM3-2T/BCM: load_watts or tdp_watts, Samsung M321R4GA3BB6-CQK 32GB DDR5-4800 RDIMM: load_watts or tdp_watts, ARCTIC Freezer 4U-SP5: load_watts or tdp_watts, Phanteks Enthoo Pro 2 Server Edition: load_watts or tdp_watts, AMD EPYC 9015: idle_watts, ASRock Rack GENOAD8QM3-2T/BCM: idle_watts, Samsung M321R4GA3BB6-CQK 32GB DDR5-4800 RDIMM: idle_watts, Sapphire Radeon AI PRO R9700 32GB (32358-01-20G): idle_watts, ARCTIC Freezer 4U-SP5: idle_watts, Phanteks Enthoo Pro 2 Server Edition: idle_watts, Seasonic PRIME PX-2200: idle_watts)	$30.00 (derived) DERIVED3× Radeon AI PRO R9700 32GB (32358-01-20G) · EPYC 9015 · 8× M321R4GA3BB6-CQK 32GB DDR5-4800 RDIMM: cost per GBArithmetic on observed inputs only. No tunable assumption enters it. cost_per_gb = landed_cost / usable_memory_gb = $10,088 / 336.32 GB = $30 /GB landed_cost 10088.4 USD — rig_landed = Σ part landed_total usable_memory_gb 336.32 GB — usable memory after the OS and framework reservation (Settings → Estimator) Some part prices are older than 36 h. LOWER BOUND: 7 of 7 parts come from a retailer that does not state a delivery charge (SabrePC, Newegg, SabrePC, Newegg, ARCTIC, Newegg, Newegg), so the rig costs at least this and never less. Usable memory is below physical memory for at least one pool (OS/driver reservation; editable in Settings).	$15.76 (derived) DERIVED3× Radeon AI PRO R9700 32GB (32358-01-20G) · EPYC 9015 · 8× M321R4GA3BB6-CQK 32GB DDR5-4800 RDIMM: cost per GB/sArithmetic on observed inputs only. No tunable assumption enters it. cost_per_gbs = landed_cost / memory_bandwidth_gbs = $10,088 / 640 GB/s = $15.76 per GB/s landed_cost 10088.4 USD — rig_landed = Σ part landed_total memory_bandwidth_gbs 640 GB/s — the published bandwidth of the rig's fastest memory pool Some part prices are older than 36 h. LOWER BOUND: 7 of 7 parts come from a retailer that does not state a delivery charge (SabrePC, Newegg, SabrePC, Newegg, ARCTIC, Newegg, Newegg), so the rig costs at least this and never less. Weights are read from the fastest pool first; a rig whose weights spill into a slower pool decodes below this figure.	Unavailable — Cost per tok/s needs decode_tok_s: GLM-5.3 has no decode speed on this rig: Model needs 507.62 GB but DIY: 3× Sapphire Radeon AI PRO R9700 32GB (32358-01-20G) + AMD EPYC 9015 has 336.32 GB usable — it will not load, so there is no decode speed.
Priced from7 offers across 3 retailersoldeststale 44 h — past the 36 hour freshness windowopen the rig below for each part’s retailer, condition and route
31	3× AI TOP Radeon AI PRO R9700 32GB (GV-R9700AI TOP-32GD) · EPYC 9015 · 8× M321R4GA3BB6-CQK 32GB DDR5-4800 RDIMMDIY workstation · 7 parts · break down its parts	6 BLOCKERS	$9,803 (derived) DERIVED3× AI TOP Radeon AI PRO R9700 32GB (GV-R9700AI TOP-32GD) · EPYC 9015 · 8× M321R4GA3BB6-CQK 32GB DDR5-4800 RDIMM: landed costArithmetic on observed inputs only. No tunable assumption enters it. rig_landed = Σ part landed_total $638 + $895 + $1,828 + $5,473 + $75 + $192 + $704 = $9,803 AMD EPYC 9015 63809 ¢ — SabrePC · ship ASRock Rack GENOAD8QM3-2T/BCM 89458 ¢ — Newegg · ship Samsung M321R4GA3BB6-CQK 32GB DDR5-4800 RDIMM 182776 ¢ — SabrePC · ship GIGABYTE AI TOP Radeon AI PRO R9700 32GB (GV-R9700AI TOP-32GD) 547272 ¢ — SabrePC · ship ARCTIC Freezer 4U-SP5 7463 ¢ — ARCTIC · ship Phanteks Enthoo Pro 2 Server Edition 19191 ¢ — Newegg · ship Seasonic PRIME PX-2200 70371 ¢ — Newegg · ship Some part prices are older than 36 h. LOWER BOUND: 7 of 7 parts come from a retailer that does not state a delivery charge (SabrePC, Newegg, SabrePC, SabrePC, ARCTIC, Newegg, Newegg), so the rig costs at least this and never less.	Unavailable — Purchase + 5-year power needs the power term. Load draw at the wall is unavailable: Electrical feasibility cannot be computed exactly: no load or TDP watts for ASRock Rack GENOAD8QM3-2T/BCM, Samsung M321R4GA3BB6-CQK 32GB DDR5-4800 RDIMM, ARCTIC Freezer 4U-SP5, Phanteks Enthoo Pro 2 Server Edition. (missing: ASRock Rack GENOAD8QM3-2T/BCM: load_watts or tdp_watts, Samsung M321R4GA3BB6-CQK 32GB DDR5-4800 RDIMM: load_watts or tdp_watts, ARCTIC Freezer 4U-SP5: load_watts or tdp_watts, Phanteks Enthoo Pro 2 Server Edition: load_watts or tdp_watts, AMD EPYC 9015: idle_watts, ASRock Rack GENOAD8QM3-2T/BCM: idle_watts, Samsung M321R4GA3BB6-CQK 32GB DDR5-4800 RDIMM: idle_watts, GIGABYTE AI TOP Radeon AI PRO R9700 32GB (GV-R9700AI TOP-32GD): idle_watts, ARCTIC Freezer 4U-SP5: idle_watts, Phanteks Enthoo Pro 2 Server Edition: idle_watts, Seasonic PRIME PX-2200: idle_watts)	$29.15 (derived) DERIVED3× AI TOP Radeon AI PRO R9700 32GB (GV-R9700AI TOP-32GD) · EPYC 9015 · 8× M321R4GA3BB6-CQK 32GB DDR5-4800 RDIMM: cost per GBArithmetic on observed inputs only. No tunable assumption enters it. cost_per_gb = landed_cost / usable_memory_gb = $9,803 / 336.32 GB = $29.15 /GB landed_cost 9803.4 USD — rig_landed = Σ part landed_total usable_memory_gb 336.32 GB — usable memory after the OS and framework reservation (Settings → Estimator) Some part prices are older than 36 h. LOWER BOUND: 7 of 7 parts come from a retailer that does not state a delivery charge (SabrePC, Newegg, SabrePC, SabrePC, ARCTIC, Newegg, Newegg), so the rig costs at least this and never less. Usable memory is below physical memory for at least one pool (OS/driver reservation; editable in Settings).	$15.32 (derived) DERIVED3× AI TOP Radeon AI PRO R9700 32GB (GV-R9700AI TOP-32GD) · EPYC 9015 · 8× M321R4GA3BB6-CQK 32GB DDR5-4800 RDIMM: cost per GB/sArithmetic on observed inputs only. No tunable assumption enters it. cost_per_gbs = landed_cost / memory_bandwidth_gbs = $9,803 / 640 GB/s = $15.32 per GB/s landed_cost 9803.4 USD — rig_landed = Σ part landed_total memory_bandwidth_gbs 640 GB/s — the published bandwidth of the rig's fastest memory pool Some part prices are older than 36 h. LOWER BOUND: 7 of 7 parts come from a retailer that does not state a delivery charge (SabrePC, Newegg, SabrePC, SabrePC, ARCTIC, Newegg, Newegg), so the rig costs at least this and never less. Weights are read from the fastest pool first; a rig whose weights spill into a slower pool decodes below this figure.	Unavailable — Cost per tok/s needs decode_tok_s: GLM-5.3 has no decode speed on this rig: Model needs 507.62 GB but DIY: 3× GIGABYTE AI TOP Radeon AI PRO R9700 32GB (GV-R9700AI TOP-32GD) + AMD EPYC 9015 has 336.32 GB usable — it will not load, so there is no decode speed.
Priced from7 offers across 3 retailersoldeststale 44 h — past the 36 hour freshness windowopen the rig below for each part’s retailer, condition and route
32	1× GeForce RTX 5090 32G GAMING TRIO OC · EPYC 9015 · 8× M321R4GA3BB6-CQK 32GB DDR5-4800 RDIMMDIY workstation · 7 parts · break down its parts	5 BLOCKERS	$9,769 (derived) DERIVED1× GeForce RTX 5090 32G GAMING TRIO OC · EPYC 9015 · 8× M321R4GA3BB6-CQK 32GB DDR5-4800 RDIMM: landed costArithmetic on observed inputs only. No tunable assumption enters it. rig_landed = Σ part landed_total $638 + $895 + $1,828 + $5,438 + $75 + $192 + $704 = $9,769 AMD EPYC 9015 63809 ¢ — SabrePC · ship ASRock Rack GENOAD8QM3-2T/BCM 89458 ¢ — Newegg · ship Samsung M321R4GA3BB6-CQK 32GB DDR5-4800 RDIMM 182776 ¢ — SabrePC · ship MSI GeForce RTX 5090 32G GAMING TRIO OC 543786 ¢ — Newegg · ship ARCTIC Freezer 4U-SP5 7463 ¢ — ARCTIC · ship Phanteks Enthoo Pro 2 Server Edition 19191 ¢ — Newegg · ship Seasonic PRIME PX-2200 70371 ¢ — Newegg · ship Some part prices are older than 36 h. LOWER BOUND: 7 of 7 parts come from a retailer that does not state a delivery charge (SabrePC, Newegg, SabrePC, Newegg, ARCTIC, Newegg, Newegg), so the rig costs at least this and never less.	Unavailable — Purchase + 5-year power needs the power term. Load draw at the wall is unavailable: Electrical feasibility cannot be computed exactly: no load or TDP watts for ASRock Rack GENOAD8QM3-2T/BCM, Samsung M321R4GA3BB6-CQK 32GB DDR5-4800 RDIMM, ARCTIC Freezer 4U-SP5, Phanteks Enthoo Pro 2 Server Edition. (missing: ASRock Rack GENOAD8QM3-2T/BCM: load_watts or tdp_watts, Samsung M321R4GA3BB6-CQK 32GB DDR5-4800 RDIMM: load_watts or tdp_watts, ARCTIC Freezer 4U-SP5: load_watts or tdp_watts, Phanteks Enthoo Pro 2 Server Edition: load_watts or tdp_watts, AMD EPYC 9015: idle_watts, ASRock Rack GENOAD8QM3-2T/BCM: idle_watts, Samsung M321R4GA3BB6-CQK 32GB DDR5-4800 RDIMM: idle_watts, MSI GeForce RTX 5090 32G GAMING TRIO OC: idle_watts, ARCTIC Freezer 4U-SP5: idle_watts, Phanteks Enthoo Pro 2 Server Edition: idle_watts, Seasonic PRIME PX-2200: idle_watts)	$35.62 (derived) DERIVED1× GeForce RTX 5090 32G GAMING TRIO OC · EPYC 9015 · 8× M321R4GA3BB6-CQK 32GB DDR5-4800 RDIMM: cost per GBArithmetic on observed inputs only. No tunable assumption enters it. cost_per_gb = landed_cost / usable_memory_gb = $9,769 / 274.24 GB = $35.62 /GB landed_cost 9768.54 USD — rig_landed = Σ part landed_total usable_memory_gb 274.24 GB — usable memory after the OS and framework reservation (Settings → Estimator) Some part prices are older than 36 h. LOWER BOUND: 7 of 7 parts come from a retailer that does not state a delivery charge (SabrePC, Newegg, SabrePC, Newegg, ARCTIC, Newegg, Newegg), so the rig costs at least this and never less. Usable memory is below physical memory for at least one pool (OS/driver reservation; editable in Settings).	$5.45 (derived) DERIVED1× GeForce RTX 5090 32G GAMING TRIO OC · EPYC 9015 · 8× M321R4GA3BB6-CQK 32GB DDR5-4800 RDIMM: cost per GB/sArithmetic on observed inputs only. No tunable assumption enters it. cost_per_gbs = landed_cost / memory_bandwidth_gbs = $9,769 / 1,792 GB/s = $5.45 per GB/s landed_cost 9768.54 USD — rig_landed = Σ part landed_total memory_bandwidth_gbs 1792 GB/s — the published bandwidth of the rig's fastest memory pool Some part prices are older than 36 h. LOWER BOUND: 7 of 7 parts come from a retailer that does not state a delivery charge (SabrePC, Newegg, SabrePC, Newegg, ARCTIC, Newegg, Newegg), so the rig costs at least this and never less. Weights are read from the fastest pool first; a rig whose weights spill into a slower pool decodes below this figure.	Unavailable — Cost per tok/s needs decode_tok_s: GLM-5.3 has no decode speed on this rig: Model needs 507.62 GB but DIY: MSI GeForce RTX 5090 32G GAMING TRIO OC + AMD EPYC 9015 has 274.24 GB usable — it will not load, so there is no decode speed.
Priced from7 offers across 3 retailersoldeststale 44 h — past the 36 hour freshness windowopen the rig below for each part’s retailer, condition and route
33	3× GeForce RTX 5080 WINDFORCE OC 16G (GV-N5080WF3OC-16GD) · EPYC 9015 · 8× M321R4GA3BB6-CQK 32GB DDR5-4800 RDIMMDIY workstation · 7 parts · break down its parts	5 BLOCKERS	$9,769 (derived) DERIVED3× GeForce RTX 5080 WINDFORCE OC 16G (GV-N5080WF3OC-16GD) · EPYC 9015 · 8× M321R4GA3BB6-CQK 32GB DDR5-4800 RDIMM: landed costArithmetic on observed inputs only. No tunable assumption enters it. rig_landed = Σ part landed_total $638 + $895 + $1,828 + $5,438 + $75 + $192 + $704 = $9,769 AMD EPYC 9015 63809 ¢ — SabrePC · ship ASRock Rack GENOAD8QM3-2T/BCM 89458 ¢ — Newegg · ship Samsung M321R4GA3BB6-CQK 32GB DDR5-4800 RDIMM 182776 ¢ — SabrePC · ship GIGABYTE GeForce RTX 5080 WINDFORCE OC 16G (GV-N5080WF3OC-16GD) 543783 ¢ — Newegg · ship ARCTIC Freezer 4U-SP5 7463 ¢ — ARCTIC · ship Phanteks Enthoo Pro 2 Server Edition 19191 ¢ — Newegg · ship Seasonic PRIME PX-2200 70371 ¢ — Newegg · ship Some part prices are older than 36 h. LOWER BOUND: 7 of 7 parts come from a retailer that does not state a delivery charge (SabrePC, Newegg, SabrePC, Newegg, ARCTIC, Newegg, Newegg), so the rig costs at least this and never less.	Unavailable — Purchase + 5-year power needs the power term. Load draw at the wall is unavailable: Electrical feasibility cannot be computed exactly: no load or TDP watts for ASRock Rack GENOAD8QM3-2T/BCM, Samsung M321R4GA3BB6-CQK 32GB DDR5-4800 RDIMM, ARCTIC Freezer 4U-SP5, Phanteks Enthoo Pro 2 Server Edition. (missing: ASRock Rack GENOAD8QM3-2T/BCM: load_watts or tdp_watts, Samsung M321R4GA3BB6-CQK 32GB DDR5-4800 RDIMM: load_watts or tdp_watts, ARCTIC Freezer 4U-SP5: load_watts or tdp_watts, Phanteks Enthoo Pro 2 Server Edition: load_watts or tdp_watts, AMD EPYC 9015: idle_watts, ASRock Rack GENOAD8QM3-2T/BCM: idle_watts, Samsung M321R4GA3BB6-CQK 32GB DDR5-4800 RDIMM: idle_watts, GIGABYTE GeForce RTX 5080 WINDFORCE OC 16G (GV-N5080WF3OC-16GD): idle_watts, ARCTIC Freezer 4U-SP5: idle_watts, Phanteks Enthoo Pro 2 Server Edition: idle_watts, Seasonic PRIME PX-2200: idle_watts)	$33.71 (derived) DERIVED3× GeForce RTX 5080 WINDFORCE OC 16G (GV-N5080WF3OC-16GD) · EPYC 9015 · 8× M321R4GA3BB6-CQK 32GB DDR5-4800 RDIMM: cost per GBArithmetic on observed inputs only. No tunable assumption enters it. cost_per_gb = landed_cost / usable_memory_gb = $9,769 / 289.76 GB = $33.71 /GB landed_cost 9768.51 USD — rig_landed = Σ part landed_total usable_memory_gb 289.76 GB — usable memory after the OS and framework reservation (Settings → Estimator) Some part prices are older than 36 h. LOWER BOUND: 7 of 7 parts come from a retailer that does not state a delivery charge (SabrePC, Newegg, SabrePC, Newegg, ARCTIC, Newegg, Newegg), so the rig costs at least this and never less. Usable memory is below physical memory for at least one pool (OS/driver reservation; editable in Settings).	$10.18 (derived) DERIVED3× GeForce RTX 5080 WINDFORCE OC 16G (GV-N5080WF3OC-16GD) · EPYC 9015 · 8× M321R4GA3BB6-CQK 32GB DDR5-4800 RDIMM: cost per GB/sArithmetic on observed inputs only. No tunable assumption enters it. cost_per_gbs = landed_cost / memory_bandwidth_gbs = $9,769 / 960 GB/s = $10.18 per GB/s landed_cost 9768.51 USD — rig_landed = Σ part landed_total memory_bandwidth_gbs 960 GB/s — the published bandwidth of the rig's fastest memory pool Some part prices are older than 36 h. LOWER BOUND: 7 of 7 parts come from a retailer that does not state a delivery charge (SabrePC, Newegg, SabrePC, Newegg, ARCTIC, Newegg, Newegg), so the rig costs at least this and never less. Weights are read from the fastest pool first; a rig whose weights spill into a slower pool decodes below this figure.	Unavailable — Cost per tok/s needs decode_tok_s: GLM-5.3 has no decode speed on this rig: Model needs 507.62 GB but DIY: 3× GIGABYTE GeForce RTX 5080 WINDFORCE OC 16G (GV-N5080WF3OC-16GD) + AMD EPYC 9015 has 289.76 GB usable — it will not load, so there is no decode speed.
Priced from7 offers across 3 retailersoldeststale 44 h — past the 36 hour freshness windowopen the rig below for each part’s retailer, condition and route
34	3× Radeon AI PRO R9700 Creator 32GB (R9700 CT) · EPYC 9015 · 8× M321R4GA3BB6-CQK 32GB DDR5-4800 RDIMMDIY workstation · 7 parts · break down its parts	6 BLOCKERS	$9,769 (derived) DERIVED3× Radeon AI PRO R9700 Creator 32GB (R9700 CT) · EPYC 9015 · 8× M321R4GA3BB6-CQK 32GB DDR5-4800 RDIMM: landed costArithmetic on observed inputs only. No tunable assumption enters it. rig_landed = Σ part landed_total $638 + $895 + $1,828 + $5,438 + $75 + $192 + $704 = $9,769 AMD EPYC 9015 63809 ¢ — SabrePC · ship ASRock Rack GENOAD8QM3-2T/BCM 89458 ¢ — Newegg · ship Samsung M321R4GA3BB6-CQK 32GB DDR5-4800 RDIMM 182776 ¢ — SabrePC · ship ASRock Radeon AI PRO R9700 Creator 32GB (R9700 CT) 543783 ¢ — Newegg · ship ARCTIC Freezer 4U-SP5 7463 ¢ — ARCTIC · ship Phanteks Enthoo Pro 2 Server Edition 19191 ¢ — Newegg · ship Seasonic PRIME PX-2200 70371 ¢ — Newegg · ship Some part prices are older than 36 h. LOWER BOUND: 7 of 7 parts come from a retailer that does not state a delivery charge (SabrePC, Newegg, SabrePC, Newegg, ARCTIC, Newegg, Newegg), so the rig costs at least this and never less.	Unavailable — Purchase + 5-year power needs the power term. Load draw at the wall is unavailable: Electrical feasibility cannot be computed exactly: no load or TDP watts for ASRock Rack GENOAD8QM3-2T/BCM, Samsung M321R4GA3BB6-CQK 32GB DDR5-4800 RDIMM, ARCTIC Freezer 4U-SP5, Phanteks Enthoo Pro 2 Server Edition. (missing: ASRock Rack GENOAD8QM3-2T/BCM: load_watts or tdp_watts, Samsung M321R4GA3BB6-CQK 32GB DDR5-4800 RDIMM: load_watts or tdp_watts, ARCTIC Freezer 4U-SP5: load_watts or tdp_watts, Phanteks Enthoo Pro 2 Server Edition: load_watts or tdp_watts, AMD EPYC 9015: idle_watts, ASRock Rack GENOAD8QM3-2T/BCM: idle_watts, Samsung M321R4GA3BB6-CQK 32GB DDR5-4800 RDIMM: idle_watts, ASRock Radeon AI PRO R9700 Creator 32GB (R9700 CT): idle_watts, ARCTIC Freezer 4U-SP5: idle_watts, Phanteks Enthoo Pro 2 Server Edition: idle_watts, Seasonic PRIME PX-2200: idle_watts)	$29.05 (derived) DERIVED3× Radeon AI PRO R9700 Creator 32GB (R9700 CT) · EPYC 9015 · 8× M321R4GA3BB6-CQK 32GB DDR5-4800 RDIMM: cost per GBArithmetic on observed inputs only. No tunable assumption enters it. cost_per_gb = landed_cost / usable_memory_gb = $9,769 / 336.32 GB = $29.05 /GB landed_cost 9768.51 USD — rig_landed = Σ part landed_total usable_memory_gb 336.32 GB — usable memory after the OS and framework reservation (Settings → Estimator) Some part prices are older than 36 h. LOWER BOUND: 7 of 7 parts come from a retailer that does not state a delivery charge (SabrePC, Newegg, SabrePC, Newegg, ARCTIC, Newegg, Newegg), so the rig costs at least this and never less. Usable memory is below physical memory for at least one pool (OS/driver reservation; editable in Settings).	$15.26 (derived) DERIVED3× Radeon AI PRO R9700 Creator 32GB (R9700 CT) · EPYC 9015 · 8× M321R4GA3BB6-CQK 32GB DDR5-4800 RDIMM: cost per GB/sArithmetic on observed inputs only. No tunable assumption enters it. cost_per_gbs = landed_cost / memory_bandwidth_gbs = $9,769 / 640 GB/s = $15.26 per GB/s landed_cost 9768.51 USD — rig_landed = Σ part landed_total memory_bandwidth_gbs 640 GB/s — the published bandwidth of the rig's fastest memory pool Some part prices are older than 36 h. LOWER BOUND: 7 of 7 parts come from a retailer that does not state a delivery charge (SabrePC, Newegg, SabrePC, Newegg, ARCTIC, Newegg, Newegg), so the rig costs at least this and never less. Weights are read from the fastest pool first; a rig whose weights spill into a slower pool decodes below this figure.	Unavailable — Cost per tok/s needs decode_tok_s: GLM-5.3 has no decode speed on this rig: Model needs 507.62 GB but DIY: 3× ASRock Radeon AI PRO R9700 Creator 32GB (R9700 CT) + AMD EPYC 9015 has 336.32 GB usable — it will not load, so there is no decode speed.
Priced from7 offers across 3 retailersoldeststale 30 d — past the 36 hour freshness windowopen the rig below for each part’s retailer, condition and route
35	3× GeForce RTX 5080 16G SHADOW 3X OC · EPYC 9015 · 8× M321R4GA3BB6-CQK 32GB DDR5-4800 RDIMMDIY workstation · 7 parts · break down its parts	5 BLOCKERS	$9,737 (derived) DERIVED3× GeForce RTX 5080 16G SHADOW 3X OC · EPYC 9015 · 8× M321R4GA3BB6-CQK 32GB DDR5-4800 RDIMM: landed costArithmetic on observed inputs only. No tunable assumption enters it. rig_landed = Σ part landed_total $638 + $895 + $1,828 + $5,406 + $75 + $192 + $704 = $9,737 AMD EPYC 9015 63809 ¢ — SabrePC · ship ASRock Rack GENOAD8QM3-2T/BCM 89458 ¢ — Newegg · ship Samsung M321R4GA3BB6-CQK 32GB DDR5-4800 RDIMM 182776 ¢ — SabrePC · ship MSI GeForce RTX 5080 16G SHADOW 3X OC 540585 ¢ — Newegg · ship ARCTIC Freezer 4U-SP5 7463 ¢ — ARCTIC · ship Phanteks Enthoo Pro 2 Server Edition 19191 ¢ — Newegg · ship Seasonic PRIME PX-2200 70371 ¢ — Newegg · ship Some part prices are older than 36 h. LOWER BOUND: 7 of 7 parts come from a retailer that does not state a delivery charge (SabrePC, Newegg, SabrePC, Newegg, ARCTIC, Newegg, Newegg), so the rig costs at least this and never less.	Unavailable — Purchase + 5-year power needs the power term. Load draw at the wall is unavailable: Electrical feasibility cannot be computed exactly: no load or TDP watts for ASRock Rack GENOAD8QM3-2T/BCM, Samsung M321R4GA3BB6-CQK 32GB DDR5-4800 RDIMM, ARCTIC Freezer 4U-SP5, Phanteks Enthoo Pro 2 Server Edition. (missing: ASRock Rack GENOAD8QM3-2T/BCM: load_watts or tdp_watts, Samsung M321R4GA3BB6-CQK 32GB DDR5-4800 RDIMM: load_watts or tdp_watts, ARCTIC Freezer 4U-SP5: load_watts or tdp_watts, Phanteks Enthoo Pro 2 Server Edition: load_watts or tdp_watts, AMD EPYC 9015: idle_watts, ASRock Rack GENOAD8QM3-2T/BCM: idle_watts, Samsung M321R4GA3BB6-CQK 32GB DDR5-4800 RDIMM: idle_watts, MSI GeForce RTX 5080 16G SHADOW 3X OC: idle_watts, ARCTIC Freezer 4U-SP5: idle_watts, Phanteks Enthoo Pro 2 Server Edition: idle_watts, Seasonic PRIME PX-2200: idle_watts)	$33.60 (derived) DERIVED3× GeForce RTX 5080 16G SHADOW 3X OC · EPYC 9015 · 8× M321R4GA3BB6-CQK 32GB DDR5-4800 RDIMM: cost per GBArithmetic on observed inputs only. No tunable assumption enters it. cost_per_gb = landed_cost / usable_memory_gb = $9,737 / 289.76 GB = $33.6 /GB landed_cost 9736.53 USD — rig_landed = Σ part landed_total usable_memory_gb 289.76 GB — usable memory after the OS and framework reservation (Settings → Estimator) Some part prices are older than 36 h. LOWER BOUND: 7 of 7 parts come from a retailer that does not state a delivery charge (SabrePC, Newegg, SabrePC, Newegg, ARCTIC, Newegg, Newegg), so the rig costs at least this and never less. Usable memory is below physical memory for at least one pool (OS/driver reservation; editable in Settings).	$10.14 (derived) DERIVED3× GeForce RTX 5080 16G SHADOW 3X OC · EPYC 9015 · 8× M321R4GA3BB6-CQK 32GB DDR5-4800 RDIMM: cost per GB/sArithmetic on observed inputs only. No tunable assumption enters it. cost_per_gbs = landed_cost / memory_bandwidth_gbs = $9,737 / 960 GB/s = $10.14 per GB/s landed_cost 9736.53 USD — rig_landed = Σ part landed_total memory_bandwidth_gbs 960 GB/s — the published bandwidth of the rig's fastest memory pool Some part prices are older than 36 h. LOWER BOUND: 7 of 7 parts come from a retailer that does not state a delivery charge (SabrePC, Newegg, SabrePC, Newegg, ARCTIC, Newegg, Newegg), so the rig costs at least this and never less. Weights are read from the fastest pool first; a rig whose weights spill into a slower pool decodes below this figure.	Unavailable — Cost per tok/s needs decode_tok_s: GLM-5.3 has no decode speed on this rig: Model needs 507.62 GB but DIY: 3× MSI GeForce RTX 5080 16G SHADOW 3X OC + AMD EPYC 9015 has 289.76 GB usable — it will not load, so there is no decode speed.
Priced from7 offers across 3 retailersoldeststale 44 h — past the 36 hour freshness windowopen the rig below for each part’s retailer, condition and route
36	GIGABYTE AI TOP ATOM (NVIDIA GB10), 128GBsealed appliance · 1 part · break down its parts	5 BLOCKERS	$8,530 (derived) DERIVEDGIGABYTE AI TOP ATOM (NVIDIA GB10), 128GB: landed costArithmetic on observed inputs only. No tunable assumption enters it. rig_landed = Σ part landed_total $8,530 = $8,530 GIGABYTE AI TOP ATOM (NVIDIA GB10), 128GB 852999 ¢ — Newegg · ship LOWER BOUND: 1 of 1 parts come from a retailer that does not state a delivery charge (Newegg), so the rig costs at least this and never less.	$9,498 (estimate) ESTIMATEGIGABYTE AI TOP ATOM (NVIDIA GB10), 128GB: purchase plus five years of powerArithmetic that also depends on a tunable assumption, listed below. purchase_plus_5yr_power = purchase + five_year_power purchase_plus_5yr_power = $8,529.99 + $967.93 = $9,497.92 purchase_cents 852999 ¢ — rig_landed = Σ part landed_total five_year_power_cents 96793 ¢ — five_year_power = annual_kwh × ¢_per_kwh × 5 LOWER BOUND: 1 of 1 parts come from a retailer that does not state a delivery charge (Newegg), so the rig costs at least this and never less. Watts are at the wall (PSU conversion losses included by the caller). Load and idle draws are averages over their hours, not peaks. Duty cycle (Settings → Power): 6 h/day at load, 18 h/day idle. 365 days per year (leap days ignored, −0.07%). Flat rate for the whole period: no escalation, no time-of-use or tiered pricing, no demand charges. Rounded to the nearest cent, half-up, once from the unrounded annual cost.	$74.05 (derived) DERIVEDGIGABYTE AI TOP ATOM (NVIDIA GB10), 128GB: cost per GBArithmetic on observed inputs only. No tunable assumption enters it. cost_per_gb = landed_cost / usable_memory_gb = $8,530 / 115.2 GB = $74.05 /GB landed_cost 8529.99 USD — rig_landed = Σ part landed_total usable_memory_gb 115.2 GB — usable memory after the OS and framework reservation (Settings → Estimator) LOWER BOUND: 1 of 1 parts come from a retailer that does not state a delivery charge (Newegg), so the rig costs at least this and never less. Usable memory is below physical memory for at least one pool (OS/driver reservation; editable in Settings).	$31.25 (derived) DERIVEDGIGABYTE AI TOP ATOM (NVIDIA GB10), 128GB: cost per GB/sArithmetic on observed inputs only. No tunable assumption enters it. cost_per_gbs = landed_cost / memory_bandwidth_gbs = $8,530 / 273 GB/s = $31.25 per GB/s landed_cost 8529.99 USD — rig_landed = Σ part landed_total memory_bandwidth_gbs 273 GB/s — the published bandwidth of the rig's fastest memory pool LOWER BOUND: 1 of 1 parts come from a retailer that does not state a delivery charge (Newegg), so the rig costs at least this and never less. Weights are read from the fastest pool first; a rig whose weights spill into a slower pool decodes below this figure.	Unavailable — Cost per tok/s needs decode_tok_s: GLM-5.3 has no decode speed on this rig: Model needs 507.62 GB but GIGABYTE AI TOP ATOM (NVIDIA GB10), 128GB has 115.2 GB usable — it will not load, so there is no decode speed.
Priced fromNeweggcondition newaccess tier 3, route newegg_realtime30 hshipped
37	NVIDIA DGX Spark (GB10 Grace Blackwell), 128GBsealed appliance · 1 part · break down its parts	5 BLOCKERS	$8,423 (derived) DERIVEDNVIDIA DGX Spark (GB10 Grace Blackwell), 128GB: landed costArithmetic on observed inputs only. No tunable assumption enters it. rig_landed = Σ part landed_total $8,423 = $8,423 NVIDIA DGX Spark (GB10 Grace Blackwell), 128GB 842336 ¢ — Central Computers · ship Some part prices are older than 36 h. LOWER BOUND: 1 of 1 parts come from a retailer that does not state a delivery charge (Central Computers), so the rig costs at least this and never less.	$9,391 (estimate) ESTIMATENVIDIA DGX Spark (GB10 Grace Blackwell), 128GB: purchase plus five years of powerArithmetic that also depends on a tunable assumption, listed below. purchase_plus_5yr_power = purchase + five_year_power purchase_plus_5yr_power = $8,423.36 + $967.93 = $9,391.29 purchase_cents 842336 ¢ — rig_landed = Σ part landed_total five_year_power_cents 96793 ¢ — five_year_power = annual_kwh × ¢_per_kwh × 5 Some part prices are older than 36 h. LOWER BOUND: 1 of 1 parts come from a retailer that does not state a delivery charge (Central Computers), so the rig costs at least this and never less. Watts are at the wall (PSU conversion losses included by the caller). Load and idle draws are averages over their hours, not peaks. Duty cycle (Settings → Power): 6 h/day at load, 18 h/day idle. 365 days per year (leap days ignored, −0.07%). Flat rate for the whole period: no escalation, no time-of-use or tiered pricing, no demand charges. Rounded to the nearest cent, half-up, once from the unrounded annual cost.	$73.12 (derived) DERIVEDNVIDIA DGX Spark (GB10 Grace Blackwell), 128GB: cost per GBArithmetic on observed inputs only. No tunable assumption enters it. cost_per_gb = landed_cost / usable_memory_gb = $8,423 / 115.2 GB = $73.12 /GB landed_cost 8423.36 USD — rig_landed = Σ part landed_total usable_memory_gb 115.2 GB — usable memory after the OS and framework reservation (Settings → Estimator) Some part prices are older than 36 h. LOWER BOUND: 1 of 1 parts come from a retailer that does not state a delivery charge (Central Computers), so the rig costs at least this and never less. Usable memory is below physical memory for at least one pool (OS/driver reservation; editable in Settings).	$30.85 (derived) DERIVEDNVIDIA DGX Spark (GB10 Grace Blackwell), 128GB: cost per GB/sArithmetic on observed inputs only. No tunable assumption enters it. cost_per_gbs = landed_cost / memory_bandwidth_gbs = $8,423 / 273 GB/s = $30.85 per GB/s landed_cost 8423.36 USD — rig_landed = Σ part landed_total memory_bandwidth_gbs 273 GB/s — the published bandwidth of the rig's fastest memory pool Some part prices are older than 36 h. LOWER BOUND: 1 of 1 parts come from a retailer that does not state a delivery charge (Central Computers), so the rig costs at least this and never less. Weights are read from the fastest pool first; a rig whose weights spill into a slower pool decodes below this figure.	Unavailable — Cost per tok/s needs decode_tok_s: GLM-5.3 has no decode speed on this rig: Model needs 507.62 GB but NVIDIA DGX Spark (GB10 Grace Blackwell), 128GB has 115.2 GB usable — it will not load, so there is no decode speed.
Priced fromCentral Computerscondition newaccess tier 1, route jsonld_productstale 9 d — past the 36 hour freshness windowshipped
38	2× Radeon AI PRO R9700 32GB (32358-01-20G) · EPYC 9015 · 8× M321R4GA3BB6-CQK 32GB DDR5-4800 RDIMMDIY workstation · 7 parts · break down its parts	6 BLOCKERS	$8,169 (derived) DERIVED2× Radeon AI PRO R9700 32GB (32358-01-20G) · EPYC 9015 · 8× M321R4GA3BB6-CQK 32GB DDR5-4800 RDIMM: landed costArithmetic on observed inputs only. No tunable assumption enters it. rig_landed = Σ part landed_total $638 + $895 + $1,828 + $3,838 + $75 + $192 + $704 = $8,169 AMD EPYC 9015 63809 ¢ — SabrePC · ship ASRock Rack GENOAD8QM3-2T/BCM 89458 ¢ — Newegg · ship Samsung M321R4GA3BB6-CQK 32GB DDR5-4800 RDIMM 182776 ¢ — SabrePC · ship Sapphire Radeon AI PRO R9700 32GB (32358-01-20G) 383848 ¢ — Newegg · ship ARCTIC Freezer 4U-SP5 7463 ¢ — ARCTIC · ship Phanteks Enthoo Pro 2 Server Edition 19191 ¢ — Newegg · ship Seasonic PRIME PX-2200 70371 ¢ — Newegg · ship Some part prices are older than 36 h. LOWER BOUND: 7 of 7 parts come from a retailer that does not state a delivery charge (SabrePC, Newegg, SabrePC, Newegg, ARCTIC, Newegg, Newegg), so the rig costs at least this and never less.	Unavailable — Purchase + 5-year power needs the power term. Load draw at the wall is unavailable: Electrical feasibility cannot be computed exactly: no load or TDP watts for ASRock Rack GENOAD8QM3-2T/BCM, Samsung M321R4GA3BB6-CQK 32GB DDR5-4800 RDIMM, ARCTIC Freezer 4U-SP5, Phanteks Enthoo Pro 2 Server Edition. (missing: ASRock Rack GENOAD8QM3-2T/BCM: load_watts or tdp_watts, Samsung M321R4GA3BB6-CQK 32GB DDR5-4800 RDIMM: load_watts or tdp_watts, ARCTIC Freezer 4U-SP5: load_watts or tdp_watts, Phanteks Enthoo Pro 2 Server Edition: load_watts or tdp_watts, AMD EPYC 9015: idle_watts, ASRock Rack GENOAD8QM3-2T/BCM: idle_watts, Samsung M321R4GA3BB6-CQK 32GB DDR5-4800 RDIMM: idle_watts, Sapphire Radeon AI PRO R9700 32GB (32358-01-20G): idle_watts, ARCTIC Freezer 4U-SP5: idle_watts, Phanteks Enthoo Pro 2 Server Edition: idle_watts, Seasonic PRIME PX-2200: idle_watts)	$26.76 (derived) DERIVED2× Radeon AI PRO R9700 32GB (32358-01-20G) · EPYC 9015 · 8× M321R4GA3BB6-CQK 32GB DDR5-4800 RDIMM: cost per GBArithmetic on observed inputs only. No tunable assumption enters it. cost_per_gb = landed_cost / usable_memory_gb = $8,169 / 305.28 GB = $26.76 /GB landed_cost 8169.16 USD — rig_landed = Σ part landed_total usable_memory_gb 305.28 GB — usable memory after the OS and framework reservation (Settings → Estimator) Some part prices are older than 36 h. LOWER BOUND: 7 of 7 parts come from a retailer that does not state a delivery charge (SabrePC, Newegg, SabrePC, Newegg, ARCTIC, Newegg, Newegg), so the rig costs at least this and never less. Usable memory is below physical memory for at least one pool (OS/driver reservation; editable in Settings).	$12.76 (derived) DERIVED2× Radeon AI PRO R9700 32GB (32358-01-20G) · EPYC 9015 · 8× M321R4GA3BB6-CQK 32GB DDR5-4800 RDIMM: cost per GB/sArithmetic on observed inputs only. No tunable assumption enters it. cost_per_gbs = landed_cost / memory_bandwidth_gbs = $8,169 / 640 GB/s = $12.76 per GB/s landed_cost 8169.16 USD — rig_landed = Σ part landed_total memory_bandwidth_gbs 640 GB/s — the published bandwidth of the rig's fastest memory pool Some part prices are older than 36 h. LOWER BOUND: 7 of 7 parts come from a retailer that does not state a delivery charge (SabrePC, Newegg, SabrePC, Newegg, ARCTIC, Newegg, Newegg), so the rig costs at least this and never less. Weights are read from the fastest pool first; a rig whose weights spill into a slower pool decodes below this figure.	Unavailable — Cost per tok/s needs decode_tok_s: GLM-5.3 has no decode speed on this rig: Model needs 507.62 GB but DIY: 2× Sapphire Radeon AI PRO R9700 32GB (32358-01-20G) + AMD EPYC 9015 has 305.28 GB usable — it will not load, so there is no decode speed.
Priced from7 offers across 3 retailersoldeststale 44 h — past the 36 hour freshness windowopen the rig below for each part’s retailer, condition and route
39	HP Z2 Mini G1a (Ryzen AI Max+ PRO 395), 128GBsealed appliance · 1 part · break down its parts	6 BLOCKERS	$8,060 (derived) DERIVEDHP Z2 Mini G1a (Ryzen AI Max+ PRO 395), 128GB: landed costArithmetic on observed inputs only. No tunable assumption enters it. rig_landed = Σ part landed_total $8,060 = $8,060 HP Z2 Mini G1a (Ryzen AI Max+ PRO 395), 128GB 805978 ¢ — CDW · ship LOWER BOUND: 1 of 1 parts come from a retailer that does not state a delivery charge (CDW), so the rig costs at least this and never less.	Unavailable — Purchase + 5-year power needs the power term. Idle draw is not published for 1 of 1 part in this rig, and the duty cycle in Settings spends most of the day idle — counting an unknown idle draw as zero would understate the bill. (missing: HP Z2 Mini G1a (Ryzen AI Max+ PRO 395), 128GB: idle_watts)	$69.96 (derived) DERIVEDHP Z2 Mini G1a (Ryzen AI Max+ PRO 395), 128GB: cost per GBArithmetic on observed inputs only. No tunable assumption enters it. cost_per_gb = landed_cost / usable_memory_gb = $8,060 / 115.2 GB = $69.96 /GB landed_cost 8059.78 USD — rig_landed = Σ part landed_total usable_memory_gb 115.2 GB — usable memory after the OS and framework reservation (Settings → Estimator) LOWER BOUND: 1 of 1 parts come from a retailer that does not state a delivery charge (CDW), so the rig costs at least this and never less. Usable memory is below physical memory for at least one pool (OS/driver reservation; editable in Settings).	$31.48 (derived) DERIVEDHP Z2 Mini G1a (Ryzen AI Max+ PRO 395), 128GB: cost per GB/sArithmetic on observed inputs only. No tunable assumption enters it. cost_per_gbs = landed_cost / memory_bandwidth_gbs = $8,060 / 256 GB/s = $31.48 per GB/s landed_cost 8059.78 USD — rig_landed = Σ part landed_total memory_bandwidth_gbs 256 GB/s — the published bandwidth of the rig's fastest memory pool LOWER BOUND: 1 of 1 parts come from a retailer that does not state a delivery charge (CDW), so the rig costs at least this and never less. Weights are read from the fastest pool first; a rig whose weights spill into a slower pool decodes below this figure.	Unavailable — Cost per tok/s needs decode_tok_s: GLM-5.3 has no decode speed on this rig: Model needs 507.62 GB but HP Z2 Mini G1a (Ryzen AI Max+ PRO 395), 128GB has 115.2 GB usable — it will not load, so there is no decode speed.
Priced fromCDWcondition newaccess tier 2, route cdw_html30 hshipped
40	2× AI TOP Radeon AI PRO R9700 32GB (GV-R9700AI TOP-32GD) · EPYC 9015 · 8× M321R4GA3BB6-CQK 32GB DDR5-4800 RDIMMDIY workstation · 7 parts · break down its parts	6 BLOCKERS	$7,979 (derived) DERIVED2× AI TOP Radeon AI PRO R9700 32GB (GV-R9700AI TOP-32GD) · EPYC 9015 · 8× M321R4GA3BB6-CQK 32GB DDR5-4800 RDIMM: landed costArithmetic on observed inputs only. No tunable assumption enters it. rig_landed = Σ part landed_total $638 + $895 + $1,828 + $3,648 + $75 + $192 + $704 = $7,979 AMD EPYC 9015 63809 ¢ — SabrePC · ship ASRock Rack GENOAD8QM3-2T/BCM 89458 ¢ — Newegg · ship Samsung M321R4GA3BB6-CQK 32GB DDR5-4800 RDIMM 182776 ¢ — SabrePC · ship GIGABYTE AI TOP Radeon AI PRO R9700 32GB (GV-R9700AI TOP-32GD) 364848 ¢ — SabrePC · ship ARCTIC Freezer 4U-SP5 7463 ¢ — ARCTIC · ship Phanteks Enthoo Pro 2 Server Edition 19191 ¢ — Newegg · ship Seasonic PRIME PX-2200 70371 ¢ — Newegg · ship Some part prices are older than 36 h. LOWER BOUND: 7 of 7 parts come from a retailer that does not state a delivery charge (SabrePC, Newegg, SabrePC, SabrePC, ARCTIC, Newegg, Newegg), so the rig costs at least this and never less.	Unavailable — Purchase + 5-year power needs the power term. Load draw at the wall is unavailable: Electrical feasibility cannot be computed exactly: no load or TDP watts for ASRock Rack GENOAD8QM3-2T/BCM, Samsung M321R4GA3BB6-CQK 32GB DDR5-4800 RDIMM, ARCTIC Freezer 4U-SP5, Phanteks Enthoo Pro 2 Server Edition. (missing: ASRock Rack GENOAD8QM3-2T/BCM: load_watts or tdp_watts, Samsung M321R4GA3BB6-CQK 32GB DDR5-4800 RDIMM: load_watts or tdp_watts, ARCTIC Freezer 4U-SP5: load_watts or tdp_watts, Phanteks Enthoo Pro 2 Server Edition: load_watts or tdp_watts, AMD EPYC 9015: idle_watts, ASRock Rack GENOAD8QM3-2T/BCM: idle_watts, Samsung M321R4GA3BB6-CQK 32GB DDR5-4800 RDIMM: idle_watts, GIGABYTE AI TOP Radeon AI PRO R9700 32GB (GV-R9700AI TOP-32GD): idle_watts, ARCTIC Freezer 4U-SP5: idle_watts, Phanteks Enthoo Pro 2 Server Edition: idle_watts, Seasonic PRIME PX-2200: idle_watts)	$26.14 (derived) DERIVED2× AI TOP Radeon AI PRO R9700 32GB (GV-R9700AI TOP-32GD) · EPYC 9015 · 8× M321R4GA3BB6-CQK 32GB DDR5-4800 RDIMM: cost per GBArithmetic on observed inputs only. No tunable assumption enters it. cost_per_gb = landed_cost / usable_memory_gb = $7,979 / 305.28 GB = $26.14 /GB landed_cost 7979.16 USD — rig_landed = Σ part landed_total usable_memory_gb 305.28 GB — usable memory after the OS and framework reservation (Settings → Estimator) Some part prices are older than 36 h. LOWER BOUND: 7 of 7 parts come from a retailer that does not state a delivery charge (SabrePC, Newegg, SabrePC, SabrePC, ARCTIC, Newegg, Newegg), so the rig costs at least this and never less. Usable memory is below physical memory for at least one pool (OS/driver reservation; editable in Settings).	$12.47 (derived) DERIVED2× AI TOP Radeon AI PRO R9700 32GB (GV-R9700AI TOP-32GD) · EPYC 9015 · 8× M321R4GA3BB6-CQK 32GB DDR5-4800 RDIMM: cost per GB/sArithmetic on observed inputs only. No tunable assumption enters it. cost_per_gbs = landed_cost / memory_bandwidth_gbs = $7,979 / 640 GB/s = $12.47 per GB/s landed_cost 7979.16 USD — rig_landed = Σ part landed_total memory_bandwidth_gbs 640 GB/s — the published bandwidth of the rig's fastest memory pool Some part prices are older than 36 h. LOWER BOUND: 7 of 7 parts come from a retailer that does not state a delivery charge (SabrePC, Newegg, SabrePC, SabrePC, ARCTIC, Newegg, Newegg), so the rig costs at least this and never less. Weights are read from the fastest pool first; a rig whose weights spill into a slower pool decodes below this figure.	Unavailable — Cost per tok/s needs decode_tok_s: GLM-5.3 has no decode speed on this rig: Model needs 507.62 GB but DIY: 2× GIGABYTE AI TOP Radeon AI PRO R9700 32GB (GV-R9700AI TOP-32GD) + AMD EPYC 9015 has 305.28 GB usable — it will not load, so there is no decode speed.
Priced from7 offers across 3 retailersoldeststale 44 h — past the 36 hour freshness windowopen the rig below for each part’s retailer, condition and route
41	2× GeForce RTX 5080 WINDFORCE OC 16G (GV-N5080WF3OC-16GD) · EPYC 9015 · 8× M321R4GA3BB6-CQK 32GB DDR5-4800 RDIMMDIY workstation · 7 parts · break down its parts	5 BLOCKERS	$7,956 (derived) DERIVED2× GeForce RTX 5080 WINDFORCE OC 16G (GV-N5080WF3OC-16GD) · EPYC 9015 · 8× M321R4GA3BB6-CQK 32GB DDR5-4800 RDIMM: landed costArithmetic on observed inputs only. No tunable assumption enters it. rig_landed = Σ part landed_total $638 + $895 + $1,828 + $3,625 + $75 + $192 + $704 = $7,956 AMD EPYC 9015 63809 ¢ — SabrePC · ship ASRock Rack GENOAD8QM3-2T/BCM 89458 ¢ — Newegg · ship Samsung M321R4GA3BB6-CQK 32GB DDR5-4800 RDIMM 182776 ¢ — SabrePC · ship GIGABYTE GeForce RTX 5080 WINDFORCE OC 16G (GV-N5080WF3OC-16GD) 362522 ¢ — Newegg · ship ARCTIC Freezer 4U-SP5 7463 ¢ — ARCTIC · ship Phanteks Enthoo Pro 2 Server Edition 19191 ¢ — Newegg · ship Seasonic PRIME PX-2200 70371 ¢ — Newegg · ship Some part prices are older than 36 h. LOWER BOUND: 7 of 7 parts come from a retailer that does not state a delivery charge (SabrePC, Newegg, SabrePC, Newegg, ARCTIC, Newegg, Newegg), so the rig costs at least this and never less.	Unavailable — Purchase + 5-year power needs the power term. Load draw at the wall is unavailable: Electrical feasibility cannot be computed exactly: no load or TDP watts for ASRock Rack GENOAD8QM3-2T/BCM, Samsung M321R4GA3BB6-CQK 32GB DDR5-4800 RDIMM, ARCTIC Freezer 4U-SP5, Phanteks Enthoo Pro 2 Server Edition. (missing: ASRock Rack GENOAD8QM3-2T/BCM: load_watts or tdp_watts, Samsung M321R4GA3BB6-CQK 32GB DDR5-4800 RDIMM: load_watts or tdp_watts, ARCTIC Freezer 4U-SP5: load_watts or tdp_watts, Phanteks Enthoo Pro 2 Server Edition: load_watts or tdp_watts, AMD EPYC 9015: idle_watts, ASRock Rack GENOAD8QM3-2T/BCM: idle_watts, Samsung M321R4GA3BB6-CQK 32GB DDR5-4800 RDIMM: idle_watts, GIGABYTE GeForce RTX 5080 WINDFORCE OC 16G (GV-N5080WF3OC-16GD): idle_watts, ARCTIC Freezer 4U-SP5: idle_watts, Phanteks Enthoo Pro 2 Server Edition: idle_watts, Seasonic PRIME PX-2200: idle_watts)	$29.01 (derived) DERIVED2× GeForce RTX 5080 WINDFORCE OC 16G (GV-N5080WF3OC-16GD) · EPYC 9015 · 8× M321R4GA3BB6-CQK 32GB DDR5-4800 RDIMM: cost per GBArithmetic on observed inputs only. No tunable assumption enters it. cost_per_gb = landed_cost / usable_memory_gb = $7,956 / 274.24 GB = $29.01 /GB landed_cost 7955.9 USD — rig_landed = Σ part landed_total usable_memory_gb 274.24 GB — usable memory after the OS and framework reservation (Settings → Estimator) Some part prices are older than 36 h. LOWER BOUND: 7 of 7 parts come from a retailer that does not state a delivery charge (SabrePC, Newegg, SabrePC, Newegg, ARCTIC, Newegg, Newegg), so the rig costs at least this and never less. Usable memory is below physical memory for at least one pool (OS/driver reservation; editable in Settings).	$8.29 (derived) DERIVED2× GeForce RTX 5080 WINDFORCE OC 16G (GV-N5080WF3OC-16GD) · EPYC 9015 · 8× M321R4GA3BB6-CQK 32GB DDR5-4800 RDIMM: cost per GB/sArithmetic on observed inputs only. No tunable assumption enters it. cost_per_gbs = landed_cost / memory_bandwidth_gbs = $7,956 / 960 GB/s = $8.29 per GB/s landed_cost 7955.9 USD — rig_landed = Σ part landed_total memory_bandwidth_gbs 960 GB/s — the published bandwidth of the rig's fastest memory pool Some part prices are older than 36 h. LOWER BOUND: 7 of 7 parts come from a retailer that does not state a delivery charge (SabrePC, Newegg, SabrePC, Newegg, ARCTIC, Newegg, Newegg), so the rig costs at least this and never less. Weights are read from the fastest pool first; a rig whose weights spill into a slower pool decodes below this figure.	Unavailable — Cost per tok/s needs decode_tok_s: GLM-5.3 has no decode speed on this rig: Model needs 507.62 GB but DIY: 2× GIGABYTE GeForce RTX 5080 WINDFORCE OC 16G (GV-N5080WF3OC-16GD) + AMD EPYC 9015 has 274.24 GB usable — it will not load, so there is no decode speed.
Priced from7 offers across 3 retailersoldeststale 44 h — past the 36 hour freshness windowopen the rig below for each part’s retailer, condition and route
42	2× Radeon AI PRO R9700 Creator 32GB (R9700 CT) · EPYC 9015 · 8× M321R4GA3BB6-CQK 32GB DDR5-4800 RDIMMDIY workstation · 7 parts · break down its parts	6 BLOCKERS	$7,956 (derived) DERIVED2× Radeon AI PRO R9700 Creator 32GB (R9700 CT) · EPYC 9015 · 8× M321R4GA3BB6-CQK 32GB DDR5-4800 RDIMM: landed costArithmetic on observed inputs only. No tunable assumption enters it. rig_landed = Σ part landed_total $638 + $895 + $1,828 + $3,625 + $75 + $192 + $704 = $7,956 AMD EPYC 9015 63809 ¢ — SabrePC · ship ASRock Rack GENOAD8QM3-2T/BCM 89458 ¢ — Newegg · ship Samsung M321R4GA3BB6-CQK 32GB DDR5-4800 RDIMM 182776 ¢ — SabrePC · ship ASRock Radeon AI PRO R9700 Creator 32GB (R9700 CT) 362522 ¢ — Newegg · ship ARCTIC Freezer 4U-SP5 7463 ¢ — ARCTIC · ship Phanteks Enthoo Pro 2 Server Edition 19191 ¢ — Newegg · ship Seasonic PRIME PX-2200 70371 ¢ — Newegg · ship Some part prices are older than 36 h. LOWER BOUND: 7 of 7 parts come from a retailer that does not state a delivery charge (SabrePC, Newegg, SabrePC, Newegg, ARCTIC, Newegg, Newegg), so the rig costs at least this and never less.	Unavailable — Purchase + 5-year power needs the power term. Load draw at the wall is unavailable: Electrical feasibility cannot be computed exactly: no load or TDP watts for ASRock Rack GENOAD8QM3-2T/BCM, Samsung M321R4GA3BB6-CQK 32GB DDR5-4800 RDIMM, ARCTIC Freezer 4U-SP5, Phanteks Enthoo Pro 2 Server Edition. (missing: ASRock Rack GENOAD8QM3-2T/BCM: load_watts or tdp_watts, Samsung M321R4GA3BB6-CQK 32GB DDR5-4800 RDIMM: load_watts or tdp_watts, ARCTIC Freezer 4U-SP5: load_watts or tdp_watts, Phanteks Enthoo Pro 2 Server Edition: load_watts or tdp_watts, AMD EPYC 9015: idle_watts, ASRock Rack GENOAD8QM3-2T/BCM: idle_watts, Samsung M321R4GA3BB6-CQK 32GB DDR5-4800 RDIMM: idle_watts, ASRock Radeon AI PRO R9700 Creator 32GB (R9700 CT): idle_watts, ARCTIC Freezer 4U-SP5: idle_watts, Phanteks Enthoo Pro 2 Server Edition: idle_watts, Seasonic PRIME PX-2200: idle_watts)	$26.06 (derived) DERIVED2× Radeon AI PRO R9700 Creator 32GB (R9700 CT) · EPYC 9015 · 8× M321R4GA3BB6-CQK 32GB DDR5-4800 RDIMM: cost per GBArithmetic on observed inputs only. No tunable assumption enters it. cost_per_gb = landed_cost / usable_memory_gb = $7,956 / 305.28 GB = $26.06 /GB landed_cost 7955.9 USD — rig_landed = Σ part landed_total usable_memory_gb 305.28 GB — usable memory after the OS and framework reservation (Settings → Estimator) Some part prices are older than 36 h. LOWER BOUND: 7 of 7 parts come from a retailer that does not state a delivery charge (SabrePC, Newegg, SabrePC, Newegg, ARCTIC, Newegg, Newegg), so the rig costs at least this and never less. Usable memory is below physical memory for at least one pool (OS/driver reservation; editable in Settings).	$12.43 (derived) DERIVED2× Radeon AI PRO R9700 Creator 32GB (R9700 CT) · EPYC 9015 · 8× M321R4GA3BB6-CQK 32GB DDR5-4800 RDIMM: cost per GB/sArithmetic on observed inputs only. No tunable assumption enters it. cost_per_gbs = landed_cost / memory_bandwidth_gbs = $7,956 / 640 GB/s = $12.43 per GB/s landed_cost 7955.9 USD — rig_landed = Σ part landed_total memory_bandwidth_gbs 640 GB/s — the published bandwidth of the rig's fastest memory pool Some part prices are older than 36 h. LOWER BOUND: 7 of 7 parts come from a retailer that does not state a delivery charge (SabrePC, Newegg, SabrePC, Newegg, ARCTIC, Newegg, Newegg), so the rig costs at least this and never less. Weights are read from the fastest pool first; a rig whose weights spill into a slower pool decodes below this figure.	Unavailable — Cost per tok/s needs decode_tok_s: GLM-5.3 has no decode speed on this rig: Model needs 507.62 GB but DIY: 2× ASRock Radeon AI PRO R9700 Creator 32GB (R9700 CT) + AMD EPYC 9015 has 305.28 GB usable — it will not load, so there is no decode speed.
Priced from7 offers across 3 retailersoldeststale 30 d — past the 36 hour freshness windowopen the rig below for each part’s retailer, condition and route
43	2× GeForce RTX 5080 16G SHADOW 3X OC · EPYC 9015 · 8× M321R4GA3BB6-CQK 32GB DDR5-4800 RDIMMDIY workstation · 7 parts · break down its parts	5 BLOCKERS	$7,935 (derived) DERIVED2× GeForce RTX 5080 16G SHADOW 3X OC · EPYC 9015 · 8× M321R4GA3BB6-CQK 32GB DDR5-4800 RDIMM: landed costArithmetic on observed inputs only. No tunable assumption enters it. rig_landed = Σ part landed_total $638 + $895 + $1,828 + $3,604 + $75 + $192 + $704 = $7,935 AMD EPYC 9015 63809 ¢ — SabrePC · ship ASRock Rack GENOAD8QM3-2T/BCM 89458 ¢ — Newegg · ship Samsung M321R4GA3BB6-CQK 32GB DDR5-4800 RDIMM 182776 ¢ — SabrePC · ship MSI GeForce RTX 5080 16G SHADOW 3X OC 360390 ¢ — Newegg · ship ARCTIC Freezer 4U-SP5 7463 ¢ — ARCTIC · ship Phanteks Enthoo Pro 2 Server Edition 19191 ¢ — Newegg · ship Seasonic PRIME PX-2200 70371 ¢ — Newegg · ship Some part prices are older than 36 h. LOWER BOUND: 7 of 7 parts come from a retailer that does not state a delivery charge (SabrePC, Newegg, SabrePC, Newegg, ARCTIC, Newegg, Newegg), so the rig costs at least this and never less.	Unavailable — Purchase + 5-year power needs the power term. Load draw at the wall is unavailable: Electrical feasibility cannot be computed exactly: no load or TDP watts for ASRock Rack GENOAD8QM3-2T/BCM, Samsung M321R4GA3BB6-CQK 32GB DDR5-4800 RDIMM, ARCTIC Freezer 4U-SP5, Phanteks Enthoo Pro 2 Server Edition. (missing: ASRock Rack GENOAD8QM3-2T/BCM: load_watts or tdp_watts, Samsung M321R4GA3BB6-CQK 32GB DDR5-4800 RDIMM: load_watts or tdp_watts, ARCTIC Freezer 4U-SP5: load_watts or tdp_watts, Phanteks Enthoo Pro 2 Server Edition: load_watts or tdp_watts, AMD EPYC 9015: idle_watts, ASRock Rack GENOAD8QM3-2T/BCM: idle_watts, Samsung M321R4GA3BB6-CQK 32GB DDR5-4800 RDIMM: idle_watts, MSI GeForce RTX 5080 16G SHADOW 3X OC: idle_watts, ARCTIC Freezer 4U-SP5: idle_watts, Phanteks Enthoo Pro 2 Server Edition: idle_watts, Seasonic PRIME PX-2200: idle_watts)	$28.93 (derived) DERIVED2× GeForce RTX 5080 16G SHADOW 3X OC · EPYC 9015 · 8× M321R4GA3BB6-CQK 32GB DDR5-4800 RDIMM: cost per GBArithmetic on observed inputs only. No tunable assumption enters it. cost_per_gb = landed_cost / usable_memory_gb = $7,935 / 274.24 GB = $28.93 /GB landed_cost 7934.58 USD — rig_landed = Σ part landed_total usable_memory_gb 274.24 GB — usable memory after the OS and framework reservation (Settings → Estimator) Some part prices are older than 36 h. LOWER BOUND: 7 of 7 parts come from a retailer that does not state a delivery charge (SabrePC, Newegg, SabrePC, Newegg, ARCTIC, Newegg, Newegg), so the rig costs at least this and never less. Usable memory is below physical memory for at least one pool (OS/driver reservation; editable in Settings).	$8.27 (derived) DERIVED2× GeForce RTX 5080 16G SHADOW 3X OC · EPYC 9015 · 8× M321R4GA3BB6-CQK 32GB DDR5-4800 RDIMM: cost per GB/sArithmetic on observed inputs only. No tunable assumption enters it. cost_per_gbs = landed_cost / memory_bandwidth_gbs = $7,935 / 960 GB/s = $8.27 per GB/s landed_cost 7934.58 USD — rig_landed = Σ part landed_total memory_bandwidth_gbs 960 GB/s — the published bandwidth of the rig's fastest memory pool Some part prices are older than 36 h. LOWER BOUND: 7 of 7 parts come from a retailer that does not state a delivery charge (SabrePC, Newegg, SabrePC, Newegg, ARCTIC, Newegg, Newegg), so the rig costs at least this and never less. Weights are read from the fastest pool first; a rig whose weights spill into a slower pool decodes below this figure.	Unavailable — Cost per tok/s needs decode_tok_s: GLM-5.3 has no decode speed on this rig: Model needs 507.62 GB but DIY: 2× MSI GeForce RTX 5080 16G SHADOW 3X OC + AMD EPYC 9015 has 274.24 GB usable — it will not load, so there is no decode speed.
Priced from7 offers across 3 retailersoldeststale 44 h — past the 36 hour freshness windowopen the rig below for each part’s retailer, condition and route
44	Apple Mac Studio (2026), M5 Ultra 36-core CPU / 80-core GPU, 96GBsealed appliance · 1 part · break down its parts	5 BLOCKERS	$7,249 (derived) DERIVEDApple Mac Studio (2026), M5 Ultra 36-core CPU / 80-core GPU, 96GB: landed costArithmetic on observed inputs only. No tunable assumption enters it. rig_landed = Σ part landed_total $7,249 = $7,249 Apple Mac Studio (2026), M5 Ultra 36-core CPU / 80-core GPU, 96GB 724943 ¢ — Apple · ship LOWER BOUND: 1 of 1 parts come from a retailer that does not state a delivery charge (Apple), so the rig costs at least this and never less.	$8,444 (estimate) ESTIMATEApple Mac Studio (2026), M5 Ultra 36-core CPU / 80-core GPU, 96GB: purchase plus five years of powerArithmetic that also depends on a tunable assumption, listed below. purchase_plus_5yr_power = purchase + five_year_power purchase_plus_5yr_power = $7,249.43 + $1,194.47 = $8,443.90 purchase_cents 724943 ¢ — rig_landed = Σ part landed_total five_year_power_cents 119447 ¢ — five_year_power = annual_kwh × ¢_per_kwh × 5 LOWER BOUND: 1 of 1 parts come from a retailer that does not state a delivery charge (Apple), so the rig costs at least this and never less. Watts are at the wall (PSU conversion losses included by the caller). Load and idle draws are averages over their hours, not peaks. Duty cycle (Settings → Power): 6 h/day at load, 18 h/day idle. 365 days per year (leap days ignored, −0.07%). Flat rate for the whole period: no escalation, no time-of-use or tiered pricing, no demand charges. Rounded to the nearest cent, half-up, once from the unrounded annual cost.	$100.69 (derived) DERIVEDApple Mac Studio (2026), M5 Ultra 36-core CPU / 80-core GPU, 96GB: cost per GBArithmetic on observed inputs only. No tunable assumption enters it. cost_per_gb = landed_cost / usable_memory_gb = $7,249 / 72 GB = $100.69 /GB landed_cost 7249.43 USD — rig_landed = Σ part landed_total usable_memory_gb 72 GB — usable memory after the OS and framework reservation (Settings → Estimator) LOWER BOUND: 1 of 1 parts come from a retailer that does not state a delivery charge (Apple), so the rig costs at least this and never less. Usable memory is below physical memory for at least one pool (OS/driver reservation; editable in Settings).	$6.04 (derived) DERIVEDApple Mac Studio (2026), M5 Ultra 36-core CPU / 80-core GPU, 96GB: cost per GB/sArithmetic on observed inputs only. No tunable assumption enters it. cost_per_gbs = landed_cost / memory_bandwidth_gbs = $7,249 / 1,200 GB/s = $6.04 per GB/s landed_cost 7249.43 USD — rig_landed = Σ part landed_total memory_bandwidth_gbs 1200 GB/s — the published bandwidth of the rig's fastest memory pool LOWER BOUND: 1 of 1 parts come from a retailer that does not state a delivery charge (Apple), so the rig costs at least this and never less. Weights are read from the fastest pool first; a rig whose weights spill into a slower pool decodes below this figure.	Unavailable — Cost per tok/s needs decode_tok_s: GLM-5.3 has no decode speed on this rig: Model needs 507.62 GB but Apple Mac Studio (2026), M5 Ultra 36-core CPU / 80-core GPU, 96GB has 72 GB usable — it will not load, so there is no decode speed.
Priced fromApplecondition newaccess tier 3, route apple_store30 hshipped
45	Hybrid · Radeon AI PRO R9700 32GB (32358-01-20G) + EPYC 9015 · 8× M321R8GA0BB0-CQK 64GB DDR5-4800 RDIMMDIY workstation · 7 parts · break down its parts	5 BLOCKERS	$6,956 (derived) DERIVEDHybrid · Radeon AI PRO R9700 32GB (32358-01-20G) + EPYC 9015 · 8× M321R8GA0BB0-CQK 64GB DDR5-4800 RDIMM: landed costArithmetic on observed inputs only. No tunable assumption enters it. rig_landed = Σ part landed_total $638 + $895 + $2,534 + $1,919 + $75 + $192 + $704 = $6,956 AMD EPYC 9015 63809 ¢ — SabrePC · ship ASRock Rack GENOAD8QM3-2T/BCM 89458 ¢ — Newegg · ship Samsung M321R8GA0BB0-CQK 64GB DDR5-4800 RDIMM 253400 ¢ — SabrePC · ship Sapphire Radeon AI PRO R9700 32GB (32358-01-20G) 191924 ¢ — Newegg · ship ARCTIC Freezer 4U-SP5 7463 ¢ — ARCTIC · ship Phanteks Enthoo Pro 2 Server Edition 19191 ¢ — Newegg · ship Seasonic PRIME PX-2200 70371 ¢ — Newegg · ship Some part prices are older than 36 h. LOWER BOUND: 7 of 7 parts come from a retailer that does not state a delivery charge (SabrePC, Newegg, SabrePC, Newegg, ARCTIC, Newegg, Newegg), so the rig costs at least this and never less.	Unavailable — Purchase + 5-year power needs the power term. Load draw at the wall is unavailable: Electrical feasibility cannot be computed exactly: no load or TDP watts for ASRock Rack GENOAD8QM3-2T/BCM, Samsung M321R8GA0BB0-CQK 64GB DDR5-4800 RDIMM, ARCTIC Freezer 4U-SP5, Phanteks Enthoo Pro 2 Server Edition. (missing: ASRock Rack GENOAD8QM3-2T/BCM: load_watts or tdp_watts, Samsung M321R8GA0BB0-CQK 64GB DDR5-4800 RDIMM: load_watts or tdp_watts, ARCTIC Freezer 4U-SP5: load_watts or tdp_watts, Phanteks Enthoo Pro 2 Server Edition: load_watts or tdp_watts, AMD EPYC 9015: idle_watts, ASRock Rack GENOAD8QM3-2T/BCM: idle_watts, Samsung M321R8GA0BB0-CQK 64GB DDR5-4800 RDIMM: idle_watts, Sapphire Radeon AI PRO R9700 32GB (32358-01-20G): idle_watts, ARCTIC Freezer 4U-SP5: idle_watts, Phanteks Enthoo Pro 2 Server Edition: idle_watts, Seasonic PRIME PX-2200: idle_watts)	$13.44 (derived) DERIVEDHybrid · Radeon AI PRO R9700 32GB (32358-01-20G) + EPYC 9015 · 8× M321R8GA0BB0-CQK 64GB DDR5-4800 RDIMM: cost per GBArithmetic on observed inputs only. No tunable assumption enters it. cost_per_gb = landed_cost / usable_memory_gb = $6,956 / 517.44 GB = $13.44 /GB landed_cost 6956.16 USD — rig_landed = Σ part landed_total usable_memory_gb 517.44 GB — usable memory after the OS and framework reservation (Settings → Estimator) Some part prices are older than 36 h. LOWER BOUND: 7 of 7 parts come from a retailer that does not state a delivery charge (SabrePC, Newegg, SabrePC, Newegg, ARCTIC, Newegg, Newegg), so the rig costs at least this and never less. Usable memory is below physical memory for at least one pool (OS/driver reservation; editable in Settings).	$10.87 (derived) DERIVEDHybrid · Radeon AI PRO R9700 32GB (32358-01-20G) + EPYC 9015 · 8× M321R8GA0BB0-CQK 64GB DDR5-4800 RDIMM: cost per GB/sArithmetic on observed inputs only. No tunable assumption enters it. cost_per_gbs = landed_cost / memory_bandwidth_gbs = $6,956 / 640 GB/s = $10.87 per GB/s landed_cost 6956.16 USD — rig_landed = Σ part landed_total memory_bandwidth_gbs 640 GB/s — the published bandwidth of the rig's fastest memory pool Some part prices are older than 36 h. LOWER BOUND: 7 of 7 parts come from a retailer that does not state a delivery charge (SabrePC, Newegg, SabrePC, Newegg, ARCTIC, Newegg, Newegg), so the rig costs at least this and never less. Weights are read from the fastest pool first; a rig whose weights spill into a slower pool decodes below this figure.	$1,190 (estimate) ESTIMATEHybrid · Radeon AI PRO R9700 32GB (32358-01-20G) + EPYC 9015 · 8× M321R8GA0BB0-CQK 64GB DDR5-4800 RDIMM: cost per tok/sArithmetic that also depends on a tunable assumption, listed below. cost_per_tok_s = landed_cost / decode_tok_s = $6,956 / 5.85 tok/s = $1,189.92 per tok/s landed_cost 6956.16 USD — rig_landed = Σ part landed_total decode_tok_s 5.85 tok/s — GLM-5.3 decoding on this rig at the Plan's context Some part prices are older than 36 h. LOWER BOUND: 7 of 7 parts come from a retailer that does not state a delivery charge (SabrePC, Newegg, SabrePC, Newegg, ARCTIC, Newegg, Newegg), so the rig costs at least this and never less. MoE: only active parameters are read per token (SPEC §3). Effective bits include embeddings and higher-precision tensors in mixed quants. Assumes active weights are quantized like the file average. KV cache is re-read on every token (DECISIONS D-008). Pools are read sequentially per token, so per-pool times add (harmonic combination). Per-token overhead on a split placement is DERIVED, not measured: each pool's own single-pool figure is charged in proportion to the share of each token it serves, because the fixed cost is per operation and the split puts that share of the operations on each device. Calibrated with that pool serving the whole model (D-013): GPU VRAM (1.58 ms), System RAM (44.61 ms). Batch size 1 (single user, interactive).
Priced from7 offers across 3 retailersoldeststale 44 h — past the 36 hour freshness windowopen the rig below for each part’s retailer, condition and route
46	Acer Veriton VGN100-UD11 (NVIDIA GB10), 128GBsealed appliance · 1 part · break down its parts	5 BLOCKERS	$6,951 (derived) DERIVEDAcer Veriton VGN100-UD11 (NVIDIA GB10), 128GB: landed costArithmetic on observed inputs only. No tunable assumption enters it. rig_landed = Σ part landed_total $6,951 = $6,951 Acer Veriton VGN100-UD11 (NVIDIA GB10), 128GB 695117 ¢ — Newegg · ship LOWER BOUND: 1 of 1 parts come from a retailer that does not state a delivery charge (Newegg), so the rig costs at least this and never less.	$7,919 (estimate) ESTIMATEAcer Veriton VGN100-UD11 (NVIDIA GB10), 128GB: purchase plus five years of powerArithmetic that also depends on a tunable assumption, listed below. purchase_plus_5yr_power = purchase + five_year_power purchase_plus_5yr_power = $6,951.17 + $967.93 = $7,919.10 purchase_cents 695117 ¢ — rig_landed = Σ part landed_total five_year_power_cents 96793 ¢ — five_year_power = annual_kwh × ¢_per_kwh × 5 LOWER BOUND: 1 of 1 parts come from a retailer that does not state a delivery charge (Newegg), so the rig costs at least this and never less. Watts are at the wall (PSU conversion losses included by the caller). Load and idle draws are averages over their hours, not peaks. Duty cycle (Settings → Power): 6 h/day at load, 18 h/day idle. 365 days per year (leap days ignored, −0.07%). Flat rate for the whole period: no escalation, no time-of-use or tiered pricing, no demand charges. Rounded to the nearest cent, half-up, once from the unrounded annual cost.	$60.34 (derived) DERIVEDAcer Veriton VGN100-UD11 (NVIDIA GB10), 128GB: cost per GBArithmetic on observed inputs only. No tunable assumption enters it. cost_per_gb = landed_cost / usable_memory_gb = $6,951 / 115.2 GB = $60.34 /GB landed_cost 6951.17 USD — rig_landed = Σ part landed_total usable_memory_gb 115.2 GB — usable memory after the OS and framework reservation (Settings → Estimator) LOWER BOUND: 1 of 1 parts come from a retailer that does not state a delivery charge (Newegg), so the rig costs at least this and never less. Usable memory is below physical memory for at least one pool (OS/driver reservation; editable in Settings).	$25.46 (derived) DERIVEDAcer Veriton VGN100-UD11 (NVIDIA GB10), 128GB: cost per GB/sArithmetic on observed inputs only. No tunable assumption enters it. cost_per_gbs = landed_cost / memory_bandwidth_gbs = $6,951 / 273 GB/s = $25.46 per GB/s landed_cost 6951.17 USD — rig_landed = Σ part landed_total memory_bandwidth_gbs 273 GB/s — the published bandwidth of the rig's fastest memory pool LOWER BOUND: 1 of 1 parts come from a retailer that does not state a delivery charge (Newegg), so the rig costs at least this and never less. Weights are read from the fastest pool first; a rig whose weights spill into a slower pool decodes below this figure.	Unavailable — Cost per tok/s needs decode_tok_s: GLM-5.3 has no decode speed on this rig: Model needs 507.62 GB but Acer Veriton VGN100-UD11 (NVIDIA GB10), 128GB has 115.2 GB usable — it will not load, so there is no decode speed.
Priced fromNeweggcondition newaccess tier 3, route newegg_realtime30 hshipped
47	Hybrid · AI TOP Radeon AI PRO R9700 32GB (GV-R9700AI TOP-32GD) + EPYC 9015 · 8× M321R8GA0BB0-CQK 64GB DDR5-4800 RDIMMDIY workstation · 7 parts · break down its parts	5 BLOCKERS	$6,861 (derived) DERIVEDHybrid · AI TOP Radeon AI PRO R9700 32GB (GV-R9700AI TOP-32GD) + EPYC 9015 · 8× M321R8GA0BB0-CQK 64GB DDR5-4800 RDIMM: landed costArithmetic on observed inputs only. No tunable assumption enters it. rig_landed = Σ part landed_total $638 + $895 + $2,534 + $1,824 + $75 + $192 + $704 = $6,861 AMD EPYC 9015 63809 ¢ — SabrePC · ship ASRock Rack GENOAD8QM3-2T/BCM 89458 ¢ — Newegg · ship Samsung M321R8GA0BB0-CQK 64GB DDR5-4800 RDIMM 253400 ¢ — SabrePC · ship GIGABYTE AI TOP Radeon AI PRO R9700 32GB (GV-R9700AI TOP-32GD) 182424 ¢ — SabrePC · ship ARCTIC Freezer 4U-SP5 7463 ¢ — ARCTIC · ship Phanteks Enthoo Pro 2 Server Edition 19191 ¢ — Newegg · ship Seasonic PRIME PX-2200 70371 ¢ — Newegg · ship Some part prices are older than 36 h. LOWER BOUND: 7 of 7 parts come from a retailer that does not state a delivery charge (SabrePC, Newegg, SabrePC, SabrePC, ARCTIC, Newegg, Newegg), so the rig costs at least this and never less.	Unavailable — Purchase + 5-year power needs the power term. Load draw at the wall is unavailable: Electrical feasibility cannot be computed exactly: no load or TDP watts for ASRock Rack GENOAD8QM3-2T/BCM, Samsung M321R8GA0BB0-CQK 64GB DDR5-4800 RDIMM, ARCTIC Freezer 4U-SP5, Phanteks Enthoo Pro 2 Server Edition. (missing: ASRock Rack GENOAD8QM3-2T/BCM: load_watts or tdp_watts, Samsung M321R8GA0BB0-CQK 64GB DDR5-4800 RDIMM: load_watts or tdp_watts, ARCTIC Freezer 4U-SP5: load_watts or tdp_watts, Phanteks Enthoo Pro 2 Server Edition: load_watts or tdp_watts, AMD EPYC 9015: idle_watts, ASRock Rack GENOAD8QM3-2T/BCM: idle_watts, Samsung M321R8GA0BB0-CQK 64GB DDR5-4800 RDIMM: idle_watts, GIGABYTE AI TOP Radeon AI PRO R9700 32GB (GV-R9700AI TOP-32GD): idle_watts, ARCTIC Freezer 4U-SP5: idle_watts, Phanteks Enthoo Pro 2 Server Edition: idle_watts, Seasonic PRIME PX-2200: idle_watts)	$13.26 (derived) DERIVEDHybrid · AI TOP Radeon AI PRO R9700 32GB (GV-R9700AI TOP-32GD) + EPYC 9015 · 8× M321R8GA0BB0-CQK 64GB DDR5-4800 RDIMM: cost per GBArithmetic on observed inputs only. No tunable assumption enters it. cost_per_gb = landed_cost / usable_memory_gb = $6,861 / 517.44 GB = $13.26 /GB landed_cost 6861.16 USD — rig_landed = Σ part landed_total usable_memory_gb 517.44 GB — usable memory after the OS and framework reservation (Settings → Estimator) Some part prices are older than 36 h. LOWER BOUND: 7 of 7 parts come from a retailer that does not state a delivery charge (SabrePC, Newegg, SabrePC, SabrePC, ARCTIC, Newegg, Newegg), so the rig costs at least this and never less. Usable memory is below physical memory for at least one pool (OS/driver reservation; editable in Settings).	$10.72 (derived) DERIVEDHybrid · AI TOP Radeon AI PRO R9700 32GB (GV-R9700AI TOP-32GD) + EPYC 9015 · 8× M321R8GA0BB0-CQK 64GB DDR5-4800 RDIMM: cost per GB/sArithmetic on observed inputs only. No tunable assumption enters it. cost_per_gbs = landed_cost / memory_bandwidth_gbs = $6,861 / 640 GB/s = $10.72 per GB/s landed_cost 6861.16 USD — rig_landed = Σ part landed_total memory_bandwidth_gbs 640 GB/s — the published bandwidth of the rig's fastest memory pool Some part prices are older than 36 h. LOWER BOUND: 7 of 7 parts come from a retailer that does not state a delivery charge (SabrePC, Newegg, SabrePC, SabrePC, ARCTIC, Newegg, Newegg), so the rig costs at least this and never less. Weights are read from the fastest pool first; a rig whose weights spill into a slower pool decodes below this figure.	$1,174 (estimate) ESTIMATEHybrid · AI TOP Radeon AI PRO R9700 32GB (GV-R9700AI TOP-32GD) + EPYC 9015 · 8× M321R8GA0BB0-CQK 64GB DDR5-4800 RDIMM: cost per tok/sArithmetic that also depends on a tunable assumption, listed below. cost_per_tok_s = landed_cost / decode_tok_s = $6,861 / 5.85 tok/s = $1,173.67 per tok/s landed_cost 6861.16 USD — rig_landed = Σ part landed_total decode_tok_s 5.85 tok/s — GLM-5.3 decoding on this rig at the Plan's context Some part prices are older than 36 h. LOWER BOUND: 7 of 7 parts come from a retailer that does not state a delivery charge (SabrePC, Newegg, SabrePC, SabrePC, ARCTIC, Newegg, Newegg), so the rig costs at least this and never less. MoE: only active parameters are read per token (SPEC §3). Effective bits include embeddings and higher-precision tensors in mixed quants. Assumes active weights are quantized like the file average. KV cache is re-read on every token (DECISIONS D-008). Pools are read sequentially per token, so per-pool times add (harmonic combination). Per-token overhead on a split placement is DERIVED, not measured: each pool's own single-pool figure is charged in proportion to the share of each token it serves, because the fixed cost is per operation and the split puts that share of the operations on each device. Calibrated with that pool serving the whole model (D-013): GPU VRAM (1.58 ms), System RAM (44.61 ms). Batch size 1 (single user, interactive).
Priced from7 offers across 3 retailersoldeststale 44 h — past the 36 hour freshness windowopen the rig below for each part’s retailer, condition and route
48	Hybrid · GeForce RTX 5080 WINDFORCE OC 16G (GV-N5080WF3OC-16GD) + EPYC 9015 · 8× M321R8GA0BB0-CQK 64GB DDR5-4800 RDIMMDIY workstation · 7 parts · break down its parts	5 BLOCKERS	$6,850 (derived) DERIVEDHybrid · GeForce RTX 5080 WINDFORCE OC 16G (GV-N5080WF3OC-16GD) + EPYC 9015 · 8× M321R8GA0BB0-CQK 64GB DDR5-4800 RDIMM: landed costArithmetic on observed inputs only. No tunable assumption enters it. rig_landed = Σ part landed_total $638 + $895 + $2,534 + $1,813 + $75 + $192 + $704 = $6,850 AMD EPYC 9015 63809 ¢ — SabrePC · ship ASRock Rack GENOAD8QM3-2T/BCM 89458 ¢ — Newegg · ship Samsung M321R8GA0BB0-CQK 64GB DDR5-4800 RDIMM 253400 ¢ — SabrePC · ship GIGABYTE GeForce RTX 5080 WINDFORCE OC 16G (GV-N5080WF3OC-16GD) 181261 ¢ — Newegg · ship ARCTIC Freezer 4U-SP5 7463 ¢ — ARCTIC · ship Phanteks Enthoo Pro 2 Server Edition 19191 ¢ — Newegg · ship Seasonic PRIME PX-2200 70371 ¢ — Newegg · ship Some part prices are older than 36 h. LOWER BOUND: 7 of 7 parts come from a retailer that does not state a delivery charge (SabrePC, Newegg, SabrePC, Newegg, ARCTIC, Newegg, Newegg), so the rig costs at least this and never less.	Unavailable — Purchase + 5-year power needs the power term. Load draw at the wall is unavailable: Electrical feasibility cannot be computed exactly: no load or TDP watts for ASRock Rack GENOAD8QM3-2T/BCM, Samsung M321R8GA0BB0-CQK 64GB DDR5-4800 RDIMM, ARCTIC Freezer 4U-SP5, Phanteks Enthoo Pro 2 Server Edition. (missing: ASRock Rack GENOAD8QM3-2T/BCM: load_watts or tdp_watts, Samsung M321R8GA0BB0-CQK 64GB DDR5-4800 RDIMM: load_watts or tdp_watts, ARCTIC Freezer 4U-SP5: load_watts or tdp_watts, Phanteks Enthoo Pro 2 Server Edition: load_watts or tdp_watts, AMD EPYC 9015: idle_watts, ASRock Rack GENOAD8QM3-2T/BCM: idle_watts, Samsung M321R8GA0BB0-CQK 64GB DDR5-4800 RDIMM: idle_watts, GIGABYTE GeForce RTX 5080 WINDFORCE OC 16G (GV-N5080WF3OC-16GD): idle_watts, ARCTIC Freezer 4U-SP5: idle_watts, Phanteks Enthoo Pro 2 Server Edition: idle_watts, Seasonic PRIME PX-2200: idle_watts)	$13.65 (derived) DERIVEDHybrid · GeForce RTX 5080 WINDFORCE OC 16G (GV-N5080WF3OC-16GD) + EPYC 9015 · 8× M321R8GA0BB0-CQK 64GB DDR5-4800 RDIMM: cost per GBArithmetic on observed inputs only. No tunable assumption enters it. cost_per_gb = landed_cost / usable_memory_gb = $6,850 / 501.92 GB = $13.65 /GB landed_cost 6849.53 USD — rig_landed = Σ part landed_total usable_memory_gb 501.92 GB — usable memory after the OS and framework reservation (Settings → Estimator) Some part prices are older than 36 h. LOWER BOUND: 7 of 7 parts come from a retailer that does not state a delivery charge (SabrePC, Newegg, SabrePC, Newegg, ARCTIC, Newegg, Newegg), so the rig costs at least this and never less. Usable memory is below physical memory for at least one pool (OS/driver reservation; editable in Settings).	$7.13 (derived) DERIVEDHybrid · GeForce RTX 5080 WINDFORCE OC 16G (GV-N5080WF3OC-16GD) + EPYC 9015 · 8× M321R8GA0BB0-CQK 64GB DDR5-4800 RDIMM: cost per GB/sArithmetic on observed inputs only. No tunable assumption enters it. cost_per_gbs = landed_cost / memory_bandwidth_gbs = $6,850 / 960 GB/s = $7.13 per GB/s landed_cost 6849.53 USD — rig_landed = Σ part landed_total memory_bandwidth_gbs 960 GB/s — the published bandwidth of the rig's fastest memory pool Some part prices are older than 36 h. LOWER BOUND: 7 of 7 parts come from a retailer that does not state a delivery charge (SabrePC, Newegg, SabrePC, Newegg, ARCTIC, Newegg, Newegg), so the rig costs at least this and never less. Weights are read from the fastest pool first; a rig whose weights spill into a slower pool decodes below this figure.	Unavailable — Cost per tok/s needs decode_tok_s: GLM-5.3 has no decode speed on this rig: Model needs 507.62 GB but DIY: GIGABYTE GeForce RTX 5080 WINDFORCE OC 16G (GV-N5080WF3OC-16GD) + AMD EPYC 9015 has 501.92 GB usable — it will not load, so there is no decode speed.
Priced from7 offers across 3 retailersoldeststale 44 h — past the 36 hour freshness windowopen the rig below for each part’s retailer, condition and route
49	Hybrid · Radeon AI PRO R9700 Creator 32GB (R9700 CT) + EPYC 9015 · 8× M321R8GA0BB0-CQK 64GB DDR5-4800 RDIMMDIY workstation · 7 parts · break down its parts	5 BLOCKERS	$6,850 (derived) DERIVEDHybrid · Radeon AI PRO R9700 Creator 32GB (R9700 CT) + EPYC 9015 · 8× M321R8GA0BB0-CQK 64GB DDR5-4800 RDIMM: landed costArithmetic on observed inputs only. No tunable assumption enters it. rig_landed = Σ part landed_total $638 + $895 + $2,534 + $1,813 + $75 + $192 + $704 = $6,850 AMD EPYC 9015 63809 ¢ — SabrePC · ship ASRock Rack GENOAD8QM3-2T/BCM 89458 ¢ — Newegg · ship Samsung M321R8GA0BB0-CQK 64GB DDR5-4800 RDIMM 253400 ¢ — SabrePC · ship ASRock Radeon AI PRO R9700 Creator 32GB (R9700 CT) 181261 ¢ — Newegg · ship ARCTIC Freezer 4U-SP5 7463 ¢ — ARCTIC · ship Phanteks Enthoo Pro 2 Server Edition 19191 ¢ — Newegg · ship Seasonic PRIME PX-2200 70371 ¢ — Newegg · ship Some part prices are older than 36 h. LOWER BOUND: 7 of 7 parts come from a retailer that does not state a delivery charge (SabrePC, Newegg, SabrePC, Newegg, ARCTIC, Newegg, Newegg), so the rig costs at least this and never less.	Unavailable — Purchase + 5-year power needs the power term. Load draw at the wall is unavailable: Electrical feasibility cannot be computed exactly: no load or TDP watts for ASRock Rack GENOAD8QM3-2T/BCM, Samsung M321R8GA0BB0-CQK 64GB DDR5-4800 RDIMM, ARCTIC Freezer 4U-SP5, Phanteks Enthoo Pro 2 Server Edition. (missing: ASRock Rack GENOAD8QM3-2T/BCM: load_watts or tdp_watts, Samsung M321R8GA0BB0-CQK 64GB DDR5-4800 RDIMM: load_watts or tdp_watts, ARCTIC Freezer 4U-SP5: load_watts or tdp_watts, Phanteks Enthoo Pro 2 Server Edition: load_watts or tdp_watts, AMD EPYC 9015: idle_watts, ASRock Rack GENOAD8QM3-2T/BCM: idle_watts, Samsung M321R8GA0BB0-CQK 64GB DDR5-4800 RDIMM: idle_watts, ASRock Radeon AI PRO R9700 Creator 32GB (R9700 CT): idle_watts, ARCTIC Freezer 4U-SP5: idle_watts, Phanteks Enthoo Pro 2 Server Edition: idle_watts, Seasonic PRIME PX-2200: idle_watts)	$13.24 (derived) DERIVEDHybrid · Radeon AI PRO R9700 Creator 32GB (R9700 CT) + EPYC 9015 · 8× M321R8GA0BB0-CQK 64GB DDR5-4800 RDIMM: cost per GBArithmetic on observed inputs only. No tunable assumption enters it. cost_per_gb = landed_cost / usable_memory_gb = $6,850 / 517.44 GB = $13.24 /GB landed_cost 6849.53 USD — rig_landed = Σ part landed_total usable_memory_gb 517.44 GB — usable memory after the OS and framework reservation (Settings → Estimator) Some part prices are older than 36 h. LOWER BOUND: 7 of 7 parts come from a retailer that does not state a delivery charge (SabrePC, Newegg, SabrePC, Newegg, ARCTIC, Newegg, Newegg), so the rig costs at least this and never less. Usable memory is below physical memory for at least one pool (OS/driver reservation; editable in Settings).	$10.70 (derived) DERIVEDHybrid · Radeon AI PRO R9700 Creator 32GB (R9700 CT) + EPYC 9015 · 8× M321R8GA0BB0-CQK 64GB DDR5-4800 RDIMM: cost per GB/sArithmetic on observed inputs only. No tunable assumption enters it. cost_per_gbs = landed_cost / memory_bandwidth_gbs = $6,850 / 640 GB/s = $10.7 per GB/s landed_cost 6849.53 USD — rig_landed = Σ part landed_total memory_bandwidth_gbs 640 GB/s — the published bandwidth of the rig's fastest memory pool Some part prices are older than 36 h. LOWER BOUND: 7 of 7 parts come from a retailer that does not state a delivery charge (SabrePC, Newegg, SabrePC, Newegg, ARCTIC, Newegg, Newegg), so the rig costs at least this and never less. Weights are read from the fastest pool first; a rig whose weights spill into a slower pool decodes below this figure.	$1,172 (estimate) ESTIMATEHybrid · Radeon AI PRO R9700 Creator 32GB (R9700 CT) + EPYC 9015 · 8× M321R8GA0BB0-CQK 64GB DDR5-4800 RDIMM: cost per tok/sArithmetic that also depends on a tunable assumption, listed below. cost_per_tok_s = landed_cost / decode_tok_s = $6,850 / 5.85 tok/s = $1,171.68 per tok/s landed_cost 6849.53 USD — rig_landed = Σ part landed_total decode_tok_s 5.85 tok/s — GLM-5.3 decoding on this rig at the Plan's context Some part prices are older than 36 h. LOWER BOUND: 7 of 7 parts come from a retailer that does not state a delivery charge (SabrePC, Newegg, SabrePC, Newegg, ARCTIC, Newegg, Newegg), so the rig costs at least this and never less. MoE: only active parameters are read per token (SPEC §3). Effective bits include embeddings and higher-precision tensors in mixed quants. Assumes active weights are quantized like the file average. KV cache is re-read on every token (DECISIONS D-008). Pools are read sequentially per token, so per-pool times add (harmonic combination). Per-token overhead on a split placement is DERIVED, not measured: each pool's own single-pool figure is charged in proportion to the share of each token it serves, because the fixed cost is per operation and the split puts that share of the operations on each device. Calibrated with that pool serving the whole model (D-013): GPU VRAM (1.58 ms), System RAM (44.61 ms). Batch size 1 (single user, interactive).
Priced from7 offers across 3 retailersoldeststale 30 d — past the 36 hour freshness windowopen the rig below for each part’s retailer, condition and route
50	Hybrid · GeForce RTX 5080 16G SHADOW 3X OC + EPYC 9015 · 8× M321R8GA0BB0-CQK 64GB DDR5-4800 RDIMMDIY workstation · 7 parts · break down its parts	5 BLOCKERS	$6,839 (derived) DERIVEDHybrid · GeForce RTX 5080 16G SHADOW 3X OC + EPYC 9015 · 8× M321R8GA0BB0-CQK 64GB DDR5-4800 RDIMM: landed costArithmetic on observed inputs only. No tunable assumption enters it. rig_landed = Σ part landed_total $638 + $895 + $2,534 + $1,802 + $75 + $192 + $704 = $6,839 AMD EPYC 9015 63809 ¢ — SabrePC · ship ASRock Rack GENOAD8QM3-2T/BCM 89458 ¢ — Newegg · ship Samsung M321R8GA0BB0-CQK 64GB DDR5-4800 RDIMM 253400 ¢ — SabrePC · ship MSI GeForce RTX 5080 16G SHADOW 3X OC 180195 ¢ — Newegg · ship ARCTIC Freezer 4U-SP5 7463 ¢ — ARCTIC · ship Phanteks Enthoo Pro 2 Server Edition 19191 ¢ — Newegg · ship Seasonic PRIME PX-2200 70371 ¢ — Newegg · ship Some part prices are older than 36 h. LOWER BOUND: 7 of 7 parts come from a retailer that does not state a delivery charge (SabrePC, Newegg, SabrePC, Newegg, ARCTIC, Newegg, Newegg), so the rig costs at least this and never less.	Unavailable — Purchase + 5-year power needs the power term. Load draw at the wall is unavailable: Electrical feasibility cannot be computed exactly: no load or TDP watts for ASRock Rack GENOAD8QM3-2T/BCM, Samsung M321R8GA0BB0-CQK 64GB DDR5-4800 RDIMM, ARCTIC Freezer 4U-SP5, Phanteks Enthoo Pro 2 Server Edition. (missing: ASRock Rack GENOAD8QM3-2T/BCM: load_watts or tdp_watts, Samsung M321R8GA0BB0-CQK 64GB DDR5-4800 RDIMM: load_watts or tdp_watts, ARCTIC Freezer 4U-SP5: load_watts or tdp_watts, Phanteks Enthoo Pro 2 Server Edition: load_watts or tdp_watts, AMD EPYC 9015: idle_watts, ASRock Rack GENOAD8QM3-2T/BCM: idle_watts, Samsung M321R8GA0BB0-CQK 64GB DDR5-4800 RDIMM: idle_watts, MSI GeForce RTX 5080 16G SHADOW 3X OC: idle_watts, ARCTIC Freezer 4U-SP5: idle_watts, Phanteks Enthoo Pro 2 Server Edition: idle_watts, Seasonic PRIME PX-2200: idle_watts)	$13.63 (derived) DERIVEDHybrid · GeForce RTX 5080 16G SHADOW 3X OC + EPYC 9015 · 8× M321R8GA0BB0-CQK 64GB DDR5-4800 RDIMM: cost per GBArithmetic on observed inputs only. No tunable assumption enters it. cost_per_gb = landed_cost / usable_memory_gb = $6,839 / 501.92 GB = $13.63 /GB landed_cost 6838.87 USD — rig_landed = Σ part landed_total usable_memory_gb 501.92 GB — usable memory after the OS and framework reservation (Settings → Estimator) Some part prices are older than 36 h. LOWER BOUND: 7 of 7 parts come from a retailer that does not state a delivery charge (SabrePC, Newegg, SabrePC, Newegg, ARCTIC, Newegg, Newegg), so the rig costs at least this and never less. Usable memory is below physical memory for at least one pool (OS/driver reservation; editable in Settings).	$7.12 (derived) DERIVEDHybrid · GeForce RTX 5080 16G SHADOW 3X OC + EPYC 9015 · 8× M321R8GA0BB0-CQK 64GB DDR5-4800 RDIMM: cost per GB/sArithmetic on observed inputs only. No tunable assumption enters it. cost_per_gbs = landed_cost / memory_bandwidth_gbs = $6,839 / 960 GB/s = $7.12 per GB/s landed_cost 6838.87 USD — rig_landed = Σ part landed_total memory_bandwidth_gbs 960 GB/s — the published bandwidth of the rig's fastest memory pool Some part prices are older than 36 h. LOWER BOUND: 7 of 7 parts come from a retailer that does not state a delivery charge (SabrePC, Newegg, SabrePC, Newegg, ARCTIC, Newegg, Newegg), so the rig costs at least this and never less. Weights are read from the fastest pool first; a rig whose weights spill into a slower pool decodes below this figure.	Unavailable — Cost per tok/s needs decode_tok_s: GLM-5.3 has no decode speed on this rig: Model needs 507.62 GB but DIY: MSI GeForce RTX 5080 16G SHADOW 3X OC + AMD EPYC 9015 has 501.92 GB usable — it will not load, so there is no decode speed.
Priced from7 offers across 3 retailersoldeststale 44 h — past the 36 hour freshness windowopen the rig below for each part’s retailer, condition and route
51	ASUS Ascent GX10 (NVIDIA GB10 Grace Blackwell), 128GBsealed appliance · 1 part · break down its parts	5 BLOCKERS	$6,396 (derived) DERIVEDASUS Ascent GX10 (NVIDIA GB10 Grace Blackwell), 128GB: landed costArithmetic on observed inputs only. No tunable assumption enters it. rig_landed = Σ part landed_total $6,396 = $6,396 ASUS Ascent GX10 (NVIDIA GB10 Grace Blackwell), 128GB 639643 ¢ — Newegg · ship LOWER BOUND: 1 of 1 parts come from a retailer that does not state a delivery charge (Newegg), so the rig costs at least this and never less.	Unavailable — Purchase + 5-year power needs the power term. Idle draw is not published for 1 of 1 part in this rig, and the duty cycle in Settings spends most of the day idle — counting an unknown idle draw as zero would understate the bill. (missing: ASUS Ascent GX10 (NVIDIA GB10 Grace Blackwell), 128GB: idle_watts)	$55.52 (derived) DERIVEDASUS Ascent GX10 (NVIDIA GB10 Grace Blackwell), 128GB: cost per GBArithmetic on observed inputs only. No tunable assumption enters it. cost_per_gb = landed_cost / usable_memory_gb = $6,396 / 115.2 GB = $55.52 /GB landed_cost 6396.43 USD — rig_landed = Σ part landed_total usable_memory_gb 115.2 GB — usable memory after the OS and framework reservation (Settings → Estimator) LOWER BOUND: 1 of 1 parts come from a retailer that does not state a delivery charge (Newegg), so the rig costs at least this and never less. Usable memory is below physical memory for at least one pool (OS/driver reservation; editable in Settings).	$23.43 (derived) DERIVEDASUS Ascent GX10 (NVIDIA GB10 Grace Blackwell), 128GB: cost per GB/sArithmetic on observed inputs only. No tunable assumption enters it. cost_per_gbs = landed_cost / memory_bandwidth_gbs = $6,396 / 273 GB/s = $23.43 per GB/s landed_cost 6396.43 USD — rig_landed = Σ part landed_total memory_bandwidth_gbs 273 GB/s — the published bandwidth of the rig's fastest memory pool LOWER BOUND: 1 of 1 parts come from a retailer that does not state a delivery charge (Newegg), so the rig costs at least this and never less. Weights are read from the fastest pool first; a rig whose weights spill into a slower pool decodes below this figure.	Unavailable — Cost per tok/s needs decode_tok_s: GLM-5.3 has no decode speed on this rig: Model needs 507.62 GB but ASUS Ascent GX10 (NVIDIA GB10 Grace Blackwell), 128GB has 115.2 GB usable — it will not load, so there is no decode speed.
Priced fromNeweggcondition newaccess tier 3, route newegg_realtime30 hshipped
52	1× Radeon AI PRO R9700 32GB (32358-01-20G) · EPYC 9015 · 8× M321R4GA3BB6-CQK 32GB DDR5-4800 RDIMMDIY workstation · 7 parts · break down its parts	6 BLOCKERS	$6,250 (derived) DERIVED1× Radeon AI PRO R9700 32GB (32358-01-20G) · EPYC 9015 · 8× M321R4GA3BB6-CQK 32GB DDR5-4800 RDIMM: landed costArithmetic on observed inputs only. No tunable assumption enters it. rig_landed = Σ part landed_total $638 + $895 + $1,828 + $1,919 + $75 + $192 + $704 = $6,250 AMD EPYC 9015 63809 ¢ — SabrePC · ship ASRock Rack GENOAD8QM3-2T/BCM 89458 ¢ — Newegg · ship Samsung M321R4GA3BB6-CQK 32GB DDR5-4800 RDIMM 182776 ¢ — SabrePC · ship Sapphire Radeon AI PRO R9700 32GB (32358-01-20G) 191924 ¢ — Newegg · ship ARCTIC Freezer 4U-SP5 7463 ¢ — ARCTIC · ship Phanteks Enthoo Pro 2 Server Edition 19191 ¢ — Newegg · ship Seasonic PRIME PX-2200 70371 ¢ — Newegg · ship Some part prices are older than 36 h. LOWER BOUND: 7 of 7 parts come from a retailer that does not state a delivery charge (SabrePC, Newegg, SabrePC, Newegg, ARCTIC, Newegg, Newegg), so the rig costs at least this and never less.	Unavailable — Purchase + 5-year power needs the power term. Load draw at the wall is unavailable: Electrical feasibility cannot be computed exactly: no load or TDP watts for ASRock Rack GENOAD8QM3-2T/BCM, Samsung M321R4GA3BB6-CQK 32GB DDR5-4800 RDIMM, ARCTIC Freezer 4U-SP5, Phanteks Enthoo Pro 2 Server Edition. (missing: ASRock Rack GENOAD8QM3-2T/BCM: load_watts or tdp_watts, Samsung M321R4GA3BB6-CQK 32GB DDR5-4800 RDIMM: load_watts or tdp_watts, ARCTIC Freezer 4U-SP5: load_watts or tdp_watts, Phanteks Enthoo Pro 2 Server Edition: load_watts or tdp_watts, AMD EPYC 9015: idle_watts, ASRock Rack GENOAD8QM3-2T/BCM: idle_watts, Samsung M321R4GA3BB6-CQK 32GB DDR5-4800 RDIMM: idle_watts, Sapphire Radeon AI PRO R9700 32GB (32358-01-20G): idle_watts, ARCTIC Freezer 4U-SP5: idle_watts, Phanteks Enthoo Pro 2 Server Edition: idle_watts, Seasonic PRIME PX-2200: idle_watts)	$22.79 (derived) DERIVED1× Radeon AI PRO R9700 32GB (32358-01-20G) · EPYC 9015 · 8× M321R4GA3BB6-CQK 32GB DDR5-4800 RDIMM: cost per GBArithmetic on observed inputs only. No tunable assumption enters it. cost_per_gb = landed_cost / usable_memory_gb = $6,250 / 274.24 GB = $22.79 /GB landed_cost 6249.92 USD — rig_landed = Σ part landed_total usable_memory_gb 274.24 GB — usable memory after the OS and framework reservation (Settings → Estimator) Some part prices are older than 36 h. LOWER BOUND: 7 of 7 parts come from a retailer that does not state a delivery charge (SabrePC, Newegg, SabrePC, Newegg, ARCTIC, Newegg, Newegg), so the rig costs at least this and never less. Usable memory is below physical memory for at least one pool (OS/driver reservation; editable in Settings).	$9.77 (derived) DERIVED1× Radeon AI PRO R9700 32GB (32358-01-20G) · EPYC 9015 · 8× M321R4GA3BB6-CQK 32GB DDR5-4800 RDIMM: cost per GB/sArithmetic on observed inputs only. No tunable assumption enters it. cost_per_gbs = landed_cost / memory_bandwidth_gbs = $6,250 / 640 GB/s = $9.77 per GB/s landed_cost 6249.92 USD — rig_landed = Σ part landed_total memory_bandwidth_gbs 640 GB/s — the published bandwidth of the rig's fastest memory pool Some part prices are older than 36 h. LOWER BOUND: 7 of 7 parts come from a retailer that does not state a delivery charge (SabrePC, Newegg, SabrePC, Newegg, ARCTIC, Newegg, Newegg), so the rig costs at least this and never less. Weights are read from the fastest pool first; a rig whose weights spill into a slower pool decodes below this figure.	Unavailable — Cost per tok/s needs decode_tok_s: GLM-5.3 has no decode speed on this rig: Model needs 507.62 GB but DIY: Sapphire Radeon AI PRO R9700 32GB (32358-01-20G) + AMD EPYC 9015 has 274.24 GB usable — it will not load, so there is no decode speed.
Priced from7 offers across 3 retailersoldeststale 44 h — past the 36 hour freshness windowopen the rig below for each part’s retailer, condition and route
53	HP ZGX Nano G1n AI Station (NVIDIA GB10), 128GBsealed appliance · 1 part · break down its parts	5 BLOCKERS	$6,237 (derived) DERIVEDHP ZGX Nano G1n AI Station (NVIDIA GB10), 128GB: landed costArithmetic on observed inputs only. No tunable assumption enters it. rig_landed = Σ part landed_total $6,237 = $6,237 HP ZGX Nano G1n AI Station (NVIDIA GB10), 128GB 623650 ¢ — HP Store · ship Some part prices are older than 36 h. LOWER BOUND: 1 of 1 parts come from a retailer that does not state a delivery charge (HP Store), so the rig costs at least this and never less.	$7,204 (estimate) ESTIMATEHP ZGX Nano G1n AI Station (NVIDIA GB10), 128GB: purchase plus five years of powerArithmetic that also depends on a tunable assumption, listed below. purchase_plus_5yr_power = purchase + five_year_power purchase_plus_5yr_power = $6,236.50 + $967.93 = $7,204.43 purchase_cents 623650 ¢ — rig_landed = Σ part landed_total five_year_power_cents 96793 ¢ — five_year_power = annual_kwh × ¢_per_kwh × 5 Some part prices are older than 36 h. LOWER BOUND: 1 of 1 parts come from a retailer that does not state a delivery charge (HP Store), so the rig costs at least this and never less. Watts are at the wall (PSU conversion losses included by the caller). Load and idle draws are averages over their hours, not peaks. Duty cycle (Settings → Power): 6 h/day at load, 18 h/day idle. 365 days per year (leap days ignored, −0.07%). Flat rate for the whole period: no escalation, no time-of-use or tiered pricing, no demand charges. Rounded to the nearest cent, half-up, once from the unrounded annual cost.	$54.14 (derived) DERIVEDHP ZGX Nano G1n AI Station (NVIDIA GB10), 128GB: cost per GBArithmetic on observed inputs only. No tunable assumption enters it. cost_per_gb = landed_cost / usable_memory_gb = $6,237 / 115.2 GB = $54.14 /GB landed_cost 6236.5 USD — rig_landed = Σ part landed_total usable_memory_gb 115.2 GB — usable memory after the OS and framework reservation (Settings → Estimator) Some part prices are older than 36 h. LOWER BOUND: 1 of 1 parts come from a retailer that does not state a delivery charge (HP Store), so the rig costs at least this and never less. Usable memory is below physical memory for at least one pool (OS/driver reservation; editable in Settings).	$22.84 (derived) DERIVEDHP ZGX Nano G1n AI Station (NVIDIA GB10), 128GB: cost per GB/sArithmetic on observed inputs only. No tunable assumption enters it. cost_per_gbs = landed_cost / memory_bandwidth_gbs = $6,237 / 273 GB/s = $22.84 per GB/s landed_cost 6236.5 USD — rig_landed = Σ part landed_total memory_bandwidth_gbs 273 GB/s — the published bandwidth of the rig's fastest memory pool Some part prices are older than 36 h. LOWER BOUND: 1 of 1 parts come from a retailer that does not state a delivery charge (HP Store), so the rig costs at least this and never less. Weights are read from the fastest pool first; a rig whose weights spill into a slower pool decodes below this figure.	Unavailable — Cost per tok/s needs decode_tok_s: GLM-5.3 has no decode speed on this rig: Model needs 507.62 GB but HP ZGX Nano G1n AI Station (NVIDIA GB10), 128GB has 115.2 GB usable — it will not load, so there is no decode speed.
Priced fromHP Storecondition newaccess tier 1, route jsonld_productstale 44 h — past the 36 hour freshness windowshipped
54	1× AI TOP Radeon AI PRO R9700 32GB (GV-R9700AI TOP-32GD) · EPYC 9015 · 8× M321R4GA3BB6-CQK 32GB DDR5-4800 RDIMMDIY workstation · 7 parts · break down its parts	6 BLOCKERS	$6,155 (derived) DERIVED1× AI TOP Radeon AI PRO R9700 32GB (GV-R9700AI TOP-32GD) · EPYC 9015 · 8× M321R4GA3BB6-CQK 32GB DDR5-4800 RDIMM: landed costArithmetic on observed inputs only. No tunable assumption enters it. rig_landed = Σ part landed_total $638 + $895 + $1,828 + $1,824 + $75 + $192 + $704 = $6,155 AMD EPYC 9015 63809 ¢ — SabrePC · ship ASRock Rack GENOAD8QM3-2T/BCM 89458 ¢ — Newegg · ship Samsung M321R4GA3BB6-CQK 32GB DDR5-4800 RDIMM 182776 ¢ — SabrePC · ship GIGABYTE AI TOP Radeon AI PRO R9700 32GB (GV-R9700AI TOP-32GD) 182424 ¢ — SabrePC · ship ARCTIC Freezer 4U-SP5 7463 ¢ — ARCTIC · ship Phanteks Enthoo Pro 2 Server Edition 19191 ¢ — Newegg · ship Seasonic PRIME PX-2200 70371 ¢ — Newegg · ship Some part prices are older than 36 h. LOWER BOUND: 7 of 7 parts come from a retailer that does not state a delivery charge (SabrePC, Newegg, SabrePC, SabrePC, ARCTIC, Newegg, Newegg), so the rig costs at least this and never less.	Unavailable — Purchase + 5-year power needs the power term. Load draw at the wall is unavailable: Electrical feasibility cannot be computed exactly: no load or TDP watts for ASRock Rack GENOAD8QM3-2T/BCM, Samsung M321R4GA3BB6-CQK 32GB DDR5-4800 RDIMM, ARCTIC Freezer 4U-SP5, Phanteks Enthoo Pro 2 Server Edition. (missing: ASRock Rack GENOAD8QM3-2T/BCM: load_watts or tdp_watts, Samsung M321R4GA3BB6-CQK 32GB DDR5-4800 RDIMM: load_watts or tdp_watts, ARCTIC Freezer 4U-SP5: load_watts or tdp_watts, Phanteks Enthoo Pro 2 Server Edition: load_watts or tdp_watts, AMD EPYC 9015: idle_watts, ASRock Rack GENOAD8QM3-2T/BCM: idle_watts, Samsung M321R4GA3BB6-CQK 32GB DDR5-4800 RDIMM: idle_watts, GIGABYTE AI TOP Radeon AI PRO R9700 32GB (GV-R9700AI TOP-32GD): idle_watts, ARCTIC Freezer 4U-SP5: idle_watts, Phanteks Enthoo Pro 2 Server Edition: idle_watts, Seasonic PRIME PX-2200: idle_watts)	$22.44 (derived) DERIVED1× AI TOP Radeon AI PRO R9700 32GB (GV-R9700AI TOP-32GD) · EPYC 9015 · 8× M321R4GA3BB6-CQK 32GB DDR5-4800 RDIMM: cost per GBArithmetic on observed inputs only. No tunable assumption enters it. cost_per_gb = landed_cost / usable_memory_gb = $6,155 / 274.24 GB = $22.44 /GB landed_cost 6154.92 USD — rig_landed = Σ part landed_total usable_memory_gb 274.24 GB — usable memory after the OS and framework reservation (Settings → Estimator) Some part prices are older than 36 h. LOWER BOUND: 7 of 7 parts come from a retailer that does not state a delivery charge (SabrePC, Newegg, SabrePC, SabrePC, ARCTIC, Newegg, Newegg), so the rig costs at least this and never less. Usable memory is below physical memory for at least one pool (OS/driver reservation; editable in Settings).	$9.62 (derived) DERIVED1× AI TOP Radeon AI PRO R9700 32GB (GV-R9700AI TOP-32GD) · EPYC 9015 · 8× M321R4GA3BB6-CQK 32GB DDR5-4800 RDIMM: cost per GB/sArithmetic on observed inputs only. No tunable assumption enters it. cost_per_gbs = landed_cost / memory_bandwidth_gbs = $6,155 / 640 GB/s = $9.62 per GB/s landed_cost 6154.92 USD — rig_landed = Σ part landed_total memory_bandwidth_gbs 640 GB/s — the published bandwidth of the rig's fastest memory pool Some part prices are older than 36 h. LOWER BOUND: 7 of 7 parts come from a retailer that does not state a delivery charge (SabrePC, Newegg, SabrePC, SabrePC, ARCTIC, Newegg, Newegg), so the rig costs at least this and never less. Weights are read from the fastest pool first; a rig whose weights spill into a slower pool decodes below this figure.	Unavailable — Cost per tok/s needs decode_tok_s: GLM-5.3 has no decode speed on this rig: Model needs 507.62 GB but DIY: GIGABYTE AI TOP Radeon AI PRO R9700 32GB (GV-R9700AI TOP-32GD) + AMD EPYC 9015 has 274.24 GB usable — it will not load, so there is no decode speed.
Priced from7 offers across 3 retailersoldeststale 44 h — past the 36 hour freshness windowopen the rig below for each part’s retailer, condition and route
55	1× GeForce RTX 5080 WINDFORCE OC 16G (GV-N5080WF3OC-16GD) · EPYC 9015 · 8× M321R4GA3BB6-CQK 32GB DDR5-4800 RDIMMDIY workstation · 7 parts · break down its parts	5 BLOCKERS	$6,143 (derived) DERIVED1× GeForce RTX 5080 WINDFORCE OC 16G (GV-N5080WF3OC-16GD) · EPYC 9015 · 8× M321R4GA3BB6-CQK 32GB DDR5-4800 RDIMM: landed costArithmetic on observed inputs only. No tunable assumption enters it. rig_landed = Σ part landed_total $638 + $895 + $1,828 + $1,813 + $75 + $192 + $704 = $6,143 AMD EPYC 9015 63809 ¢ — SabrePC · ship ASRock Rack GENOAD8QM3-2T/BCM 89458 ¢ — Newegg · ship Samsung M321R4GA3BB6-CQK 32GB DDR5-4800 RDIMM 182776 ¢ — SabrePC · ship GIGABYTE GeForce RTX 5080 WINDFORCE OC 16G (GV-N5080WF3OC-16GD) 181261 ¢ — Newegg · ship ARCTIC Freezer 4U-SP5 7463 ¢ — ARCTIC · ship Phanteks Enthoo Pro 2 Server Edition 19191 ¢ — Newegg · ship Seasonic PRIME PX-2200 70371 ¢ — Newegg · ship Some part prices are older than 36 h. LOWER BOUND: 7 of 7 parts come from a retailer that does not state a delivery charge (SabrePC, Newegg, SabrePC, Newegg, ARCTIC, Newegg, Newegg), so the rig costs at least this and never less.	Unavailable — Purchase + 5-year power needs the power term. Load draw at the wall is unavailable: Electrical feasibility cannot be computed exactly: no load or TDP watts for ASRock Rack GENOAD8QM3-2T/BCM, Samsung M321R4GA3BB6-CQK 32GB DDR5-4800 RDIMM, ARCTIC Freezer 4U-SP5, Phanteks Enthoo Pro 2 Server Edition. (missing: ASRock Rack GENOAD8QM3-2T/BCM: load_watts or tdp_watts, Samsung M321R4GA3BB6-CQK 32GB DDR5-4800 RDIMM: load_watts or tdp_watts, ARCTIC Freezer 4U-SP5: load_watts or tdp_watts, Phanteks Enthoo Pro 2 Server Edition: load_watts or tdp_watts, AMD EPYC 9015: idle_watts, ASRock Rack GENOAD8QM3-2T/BCM: idle_watts, Samsung M321R4GA3BB6-CQK 32GB DDR5-4800 RDIMM: idle_watts, GIGABYTE GeForce RTX 5080 WINDFORCE OC 16G (GV-N5080WF3OC-16GD): idle_watts, ARCTIC Freezer 4U-SP5: idle_watts, Phanteks Enthoo Pro 2 Server Edition: idle_watts, Seasonic PRIME PX-2200: idle_watts)	$23.74 (derived) DERIVED1× GeForce RTX 5080 WINDFORCE OC 16G (GV-N5080WF3OC-16GD) · EPYC 9015 · 8× M321R4GA3BB6-CQK 32GB DDR5-4800 RDIMM: cost per GBArithmetic on observed inputs only. No tunable assumption enters it. cost_per_gb = landed_cost / usable_memory_gb = $6,143 / 258.72 GB = $23.74 /GB landed_cost 6143.29 USD — rig_landed = Σ part landed_total usable_memory_gb 258.72 GB — usable memory after the OS and framework reservation (Settings → Estimator) Some part prices are older than 36 h. LOWER BOUND: 7 of 7 parts come from a retailer that does not state a delivery charge (SabrePC, Newegg, SabrePC, Newegg, ARCTIC, Newegg, Newegg), so the rig costs at least this and never less. Usable memory is below physical memory for at least one pool (OS/driver reservation; editable in Settings).	$6.40 (derived) DERIVED1× GeForce RTX 5080 WINDFORCE OC 16G (GV-N5080WF3OC-16GD) · EPYC 9015 · 8× M321R4GA3BB6-CQK 32GB DDR5-4800 RDIMM: cost per GB/sArithmetic on observed inputs only. No tunable assumption enters it. cost_per_gbs = landed_cost / memory_bandwidth_gbs = $6,143 / 960 GB/s = $6.4 per GB/s landed_cost 6143.29 USD — rig_landed = Σ part landed_total memory_bandwidth_gbs 960 GB/s — the published bandwidth of the rig's fastest memory pool Some part prices are older than 36 h. LOWER BOUND: 7 of 7 parts come from a retailer that does not state a delivery charge (SabrePC, Newegg, SabrePC, Newegg, ARCTIC, Newegg, Newegg), so the rig costs at least this and never less. Weights are read from the fastest pool first; a rig whose weights spill into a slower pool decodes below this figure.	Unavailable — Cost per tok/s needs decode_tok_s: GLM-5.3 has no decode speed on this rig: Model needs 507.62 GB but DIY: GIGABYTE GeForce RTX 5080 WINDFORCE OC 16G (GV-N5080WF3OC-16GD) + AMD EPYC 9015 has 258.72 GB usable — it will not load, so there is no decode speed.
Priced from7 offers across 3 retailersoldeststale 44 h — past the 36 hour freshness windowopen the rig below for each part’s retailer, condition and route
56	1× Radeon AI PRO R9700 Creator 32GB (R9700 CT) · EPYC 9015 · 8× M321R4GA3BB6-CQK 32GB DDR5-4800 RDIMMDIY workstation · 7 parts · break down its parts	6 BLOCKERS	$6,143 (derived) DERIVED1× Radeon AI PRO R9700 Creator 32GB (R9700 CT) · EPYC 9015 · 8× M321R4GA3BB6-CQK 32GB DDR5-4800 RDIMM: landed costArithmetic on observed inputs only. No tunable assumption enters it. rig_landed = Σ part landed_total $638 + $895 + $1,828 + $1,813 + $75 + $192 + $704 = $6,143 AMD EPYC 9015 63809 ¢ — SabrePC · ship ASRock Rack GENOAD8QM3-2T/BCM 89458 ¢ — Newegg · ship Samsung M321R4GA3BB6-CQK 32GB DDR5-4800 RDIMM 182776 ¢ — SabrePC · ship ASRock Radeon AI PRO R9700 Creator 32GB (R9700 CT) 181261 ¢ — Newegg · ship ARCTIC Freezer 4U-SP5 7463 ¢ — ARCTIC · ship Phanteks Enthoo Pro 2 Server Edition 19191 ¢ — Newegg · ship Seasonic PRIME PX-2200 70371 ¢ — Newegg · ship Some part prices are older than 36 h. LOWER BOUND: 7 of 7 parts come from a retailer that does not state a delivery charge (SabrePC, Newegg, SabrePC, Newegg, ARCTIC, Newegg, Newegg), so the rig costs at least this and never less.	Unavailable — Purchase + 5-year power needs the power term. Load draw at the wall is unavailable: Electrical feasibility cannot be computed exactly: no load or TDP watts for ASRock Rack GENOAD8QM3-2T/BCM, Samsung M321R4GA3BB6-CQK 32GB DDR5-4800 RDIMM, ARCTIC Freezer 4U-SP5, Phanteks Enthoo Pro 2 Server Edition. (missing: ASRock Rack GENOAD8QM3-2T/BCM: load_watts or tdp_watts, Samsung M321R4GA3BB6-CQK 32GB DDR5-4800 RDIMM: load_watts or tdp_watts, ARCTIC Freezer 4U-SP5: load_watts or tdp_watts, Phanteks Enthoo Pro 2 Server Edition: load_watts or tdp_watts, AMD EPYC 9015: idle_watts, ASRock Rack GENOAD8QM3-2T/BCM: idle_watts, Samsung M321R4GA3BB6-CQK 32GB DDR5-4800 RDIMM: idle_watts, ASRock Radeon AI PRO R9700 Creator 32GB (R9700 CT): idle_watts, ARCTIC Freezer 4U-SP5: idle_watts, Phanteks Enthoo Pro 2 Server Edition: idle_watts, Seasonic PRIME PX-2200: idle_watts)	$22.40 (derived) DERIVED1× Radeon AI PRO R9700 Creator 32GB (R9700 CT) · EPYC 9015 · 8× M321R4GA3BB6-CQK 32GB DDR5-4800 RDIMM: cost per GBArithmetic on observed inputs only. No tunable assumption enters it. cost_per_gb = landed_cost / usable_memory_gb = $6,143 / 274.24 GB = $22.4 /GB landed_cost 6143.29 USD — rig_landed = Σ part landed_total usable_memory_gb 274.24 GB — usable memory after the OS and framework reservation (Settings → Estimator) Some part prices are older than 36 h. LOWER BOUND: 7 of 7 parts come from a retailer that does not state a delivery charge (SabrePC, Newegg, SabrePC, Newegg, ARCTIC, Newegg, Newegg), so the rig costs at least this and never less. Usable memory is below physical memory for at least one pool (OS/driver reservation; editable in Settings).	$9.60 (derived) DERIVED1× Radeon AI PRO R9700 Creator 32GB (R9700 CT) · EPYC 9015 · 8× M321R4GA3BB6-CQK 32GB DDR5-4800 RDIMM: cost per GB/sArithmetic on observed inputs only. No tunable assumption enters it. cost_per_gbs = landed_cost / memory_bandwidth_gbs = $6,143 / 640 GB/s = $9.6 per GB/s landed_cost 6143.29 USD — rig_landed = Σ part landed_total memory_bandwidth_gbs 640 GB/s — the published bandwidth of the rig's fastest memory pool Some part prices are older than 36 h. LOWER BOUND: 7 of 7 parts come from a retailer that does not state a delivery charge (SabrePC, Newegg, SabrePC, Newegg, ARCTIC, Newegg, Newegg), so the rig costs at least this and never less. Weights are read from the fastest pool first; a rig whose weights spill into a slower pool decodes below this figure.	Unavailable — Cost per tok/s needs decode_tok_s: GLM-5.3 has no decode speed on this rig: Model needs 507.62 GB but DIY: ASRock Radeon AI PRO R9700 Creator 32GB (R9700 CT) + AMD EPYC 9015 has 274.24 GB usable — it will not load, so there is no decode speed.
Priced from7 offers across 3 retailersoldeststale 30 d — past the 36 hour freshness windowopen the rig below for each part’s retailer, condition and route
57	1× GeForce RTX 5080 16G SHADOW 3X OC · EPYC 9015 · 8× M321R4GA3BB6-CQK 32GB DDR5-4800 RDIMMDIY workstation · 7 parts · break down its parts	5 BLOCKERS	$6,133 (derived) DERIVED1× GeForce RTX 5080 16G SHADOW 3X OC · EPYC 9015 · 8× M321R4GA3BB6-CQK 32GB DDR5-4800 RDIMM: landed costArithmetic on observed inputs only. No tunable assumption enters it. rig_landed = Σ part landed_total $638 + $895 + $1,828 + $1,802 + $75 + $192 + $704 = $6,133 AMD EPYC 9015 63809 ¢ — SabrePC · ship ASRock Rack GENOAD8QM3-2T/BCM 89458 ¢ — Newegg · ship Samsung M321R4GA3BB6-CQK 32GB DDR5-4800 RDIMM 182776 ¢ — SabrePC · ship MSI GeForce RTX 5080 16G SHADOW 3X OC 180195 ¢ — Newegg · ship ARCTIC Freezer 4U-SP5 7463 ¢ — ARCTIC · ship Phanteks Enthoo Pro 2 Server Edition 19191 ¢ — Newegg · ship Seasonic PRIME PX-2200 70371 ¢ — Newegg · ship Some part prices are older than 36 h. LOWER BOUND: 7 of 7 parts come from a retailer that does not state a delivery charge (SabrePC, Newegg, SabrePC, Newegg, ARCTIC, Newegg, Newegg), so the rig costs at least this and never less.	Unavailable — Purchase + 5-year power needs the power term. Load draw at the wall is unavailable: Electrical feasibility cannot be computed exactly: no load or TDP watts for ASRock Rack GENOAD8QM3-2T/BCM, Samsung M321R4GA3BB6-CQK 32GB DDR5-4800 RDIMM, ARCTIC Freezer 4U-SP5, Phanteks Enthoo Pro 2 Server Edition. (missing: ASRock Rack GENOAD8QM3-2T/BCM: load_watts or tdp_watts, Samsung M321R4GA3BB6-CQK 32GB DDR5-4800 RDIMM: load_watts or tdp_watts, ARCTIC Freezer 4U-SP5: load_watts or tdp_watts, Phanteks Enthoo Pro 2 Server Edition: load_watts or tdp_watts, AMD EPYC 9015: idle_watts, ASRock Rack GENOAD8QM3-2T/BCM: idle_watts, Samsung M321R4GA3BB6-CQK 32GB DDR5-4800 RDIMM: idle_watts, MSI GeForce RTX 5080 16G SHADOW 3X OC: idle_watts, ARCTIC Freezer 4U-SP5: idle_watts, Phanteks Enthoo Pro 2 Server Edition: idle_watts, Seasonic PRIME PX-2200: idle_watts)	$23.70 (derived) DERIVED1× GeForce RTX 5080 16G SHADOW 3X OC · EPYC 9015 · 8× M321R4GA3BB6-CQK 32GB DDR5-4800 RDIMM: cost per GBArithmetic on observed inputs only. No tunable assumption enters it. cost_per_gb = landed_cost / usable_memory_gb = $6,133 / 258.72 GB = $23.7 /GB landed_cost 6132.63 USD — rig_landed = Σ part landed_total usable_memory_gb 258.72 GB — usable memory after the OS and framework reservation (Settings → Estimator) Some part prices are older than 36 h. LOWER BOUND: 7 of 7 parts come from a retailer that does not state a delivery charge (SabrePC, Newegg, SabrePC, Newegg, ARCTIC, Newegg, Newegg), so the rig costs at least this and never less. Usable memory is below physical memory for at least one pool (OS/driver reservation; editable in Settings).	$6.39 (derived) DERIVED1× GeForce RTX 5080 16G SHADOW 3X OC · EPYC 9015 · 8× M321R4GA3BB6-CQK 32GB DDR5-4800 RDIMM: cost per GB/sArithmetic on observed inputs only. No tunable assumption enters it. cost_per_gbs = landed_cost / memory_bandwidth_gbs = $6,133 / 960 GB/s = $6.39 per GB/s landed_cost 6132.63 USD — rig_landed = Σ part landed_total memory_bandwidth_gbs 960 GB/s — the published bandwidth of the rig's fastest memory pool Some part prices are older than 36 h. LOWER BOUND: 7 of 7 parts come from a retailer that does not state a delivery charge (SabrePC, Newegg, SabrePC, Newegg, ARCTIC, Newegg, Newegg), so the rig costs at least this and never less. Weights are read from the fastest pool first; a rig whose weights spill into a slower pool decodes below this figure.	Unavailable — Cost per tok/s needs decode_tok_s: GLM-5.3 has no decode speed on this rig: Model needs 507.62 GB but DIY: MSI GeForce RTX 5080 16G SHADOW 3X OC + AMD EPYC 9015 has 258.72 GB usable — it will not load, so there is no decode speed.
Priced from7 offers across 3 retailersoldeststale 44 h — past the 36 hour freshness windowopen the rig below for each part’s retailer, condition and route
58	Apple Mac Studio (2026), M5 Ultra 30-core CPU / 64-core GPU, 96GBsealed appliance · 1 part · break down its parts	5 BLOCKERS	$5,863 (derived) DERIVEDApple Mac Studio (2026), M5 Ultra 30-core CPU / 64-core GPU, 96GB: landed costArithmetic on observed inputs only. No tunable assumption enters it. rig_landed = Σ part landed_total $5,863 = $5,863 Apple Mac Studio (2026), M5 Ultra 30-core CPU / 64-core GPU, 96GB 586331 ¢ — Apple · ship LOWER BOUND: 1 of 1 parts come from a retailer that does not state a delivery charge (Apple), so the rig costs at least this and never less.	Unavailable — Purchase + 5-year power needs the power term. Idle draw is not published for 1 of 1 part in this rig, and the duty cycle in Settings spends most of the day idle — counting an unknown idle draw as zero would understate the bill. (missing: Apple Mac Studio (2026), M5 Ultra 30-core CPU / 64-core GPU, 96GB: idle_watts)	$81.43 (derived) DERIVEDApple Mac Studio (2026), M5 Ultra 30-core CPU / 64-core GPU, 96GB: cost per GBArithmetic on observed inputs only. No tunable assumption enters it. cost_per_gb = landed_cost / usable_memory_gb = $5,863 / 72 GB = $81.43 /GB landed_cost 5863.31 USD — rig_landed = Σ part landed_total usable_memory_gb 72 GB — usable memory after the OS and framework reservation (Settings → Estimator) LOWER BOUND: 1 of 1 parts come from a retailer that does not state a delivery charge (Apple), so the rig costs at least this and never less. Usable memory is below physical memory for at least one pool (OS/driver reservation; editable in Settings).	$4.89 (derived) DERIVEDApple Mac Studio (2026), M5 Ultra 30-core CPU / 64-core GPU, 96GB: cost per GB/sArithmetic on observed inputs only. No tunable assumption enters it. cost_per_gbs = landed_cost / memory_bandwidth_gbs = $5,863 / 1,200 GB/s = $4.89 per GB/s landed_cost 5863.31 USD — rig_landed = Σ part landed_total memory_bandwidth_gbs 1200 GB/s — the published bandwidth of the rig's fastest memory pool LOWER BOUND: 1 of 1 parts come from a retailer that does not state a delivery charge (Apple), so the rig costs at least this and never less. Weights are read from the fastest pool first; a rig whose weights spill into a slower pool decodes below this figure.	Unavailable — Cost per tok/s needs decode_tok_s: GLM-5.3 has no decode speed on this rig: Model needs 507.62 GB but Apple Mac Studio (2026), M5 Ultra 30-core CPU / 64-core GPU, 96GB has 72 GB usable — it will not load, so there is no decode speed.
Priced fromApplecondition newaccess tier 3, route apple_store30 hshipped
59	Apple Mac Studio (2026), M5 Max 18-core CPU / 40-core GPU, 128GBsealed appliance · 1 part · break down its parts	5 BLOCKERS	$5,757 (derived) DERIVEDApple Mac Studio (2026), M5 Max 18-core CPU / 40-core GPU, 128GB: landed costArithmetic on observed inputs only. No tunable assumption enters it. rig_landed = Σ part landed_total $5,757 = $5,757 Apple Mac Studio (2026), M5 Max 18-core CPU / 40-core GPU, 128GB 575668 ¢ — Apple · ship LOWER BOUND: 1 of 1 parts come from a retailer that does not state a delivery charge (Apple), so the rig costs at least this and never less.	$7,222 (estimate) ESTIMATEApple Mac Studio (2026), M5 Max 18-core CPU / 40-core GPU, 128GB: purchase plus five years of powerArithmetic that also depends on a tunable assumption, listed below. purchase_plus_5yr_power = purchase + five_year_power purchase_plus_5yr_power = $5,756.68 + $1,465.14 = $7,221.82 purchase_cents 575668 ¢ — rig_landed = Σ part landed_total five_year_power_cents 146514 ¢ — five_year_power = annual_kwh × ¢_per_kwh × 5 LOWER BOUND: 1 of 1 parts come from a retailer that does not state a delivery charge (Apple), so the rig costs at least this and never less. Watts are at the wall (PSU conversion losses included by the caller). Load and idle draws are averages over their hours, not peaks. Duty cycle (Settings → Power): 6 h/day at load, 18 h/day idle. 365 days per year (leap days ignored, −0.07%). Flat rate for the whole period: no escalation, no time-of-use or tiered pricing, no demand charges. Rounded to the nearest cent, half-up, once from the unrounded annual cost.	$59.97 (derived) DERIVEDApple Mac Studio (2026), M5 Max 18-core CPU / 40-core GPU, 128GB: cost per GBArithmetic on observed inputs only. No tunable assumption enters it. cost_per_gb = landed_cost / usable_memory_gb = $5,757 / 96 GB = $59.97 /GB landed_cost 5756.68 USD — rig_landed = Σ part landed_total usable_memory_gb 96 GB — usable memory after the OS and framework reservation (Settings → Estimator) LOWER BOUND: 1 of 1 parts come from a retailer that does not state a delivery charge (Apple), so the rig costs at least this and never less. Usable memory is below physical memory for at least one pool (OS/driver reservation; editable in Settings).	$9.38 (derived) DERIVEDApple Mac Studio (2026), M5 Max 18-core CPU / 40-core GPU, 128GB: cost per GB/sArithmetic on observed inputs only. No tunable assumption enters it. cost_per_gbs = landed_cost / memory_bandwidth_gbs = $5,757 / 614 GB/s = $9.38 per GB/s landed_cost 5756.68 USD — rig_landed = Σ part landed_total memory_bandwidth_gbs 614 GB/s — the published bandwidth of the rig's fastest memory pool LOWER BOUND: 1 of 1 parts come from a retailer that does not state a delivery charge (Apple), so the rig costs at least this and never less. Weights are read from the fastest pool first; a rig whose weights spill into a slower pool decodes below this figure.	Unavailable — Cost per tok/s needs decode_tok_s: GLM-5.3 has no decode speed on this rig: Model needs 507.62 GB but Apple Mac Studio (2026), M5 Max 18-core CPU / 40-core GPU, 128GB has 96 GB usable — it will not load, so there is no decode speed.
Priced fromApplecondition newaccess tier 3, route apple_store30 hshipped
60	MSI EdgeXpert (NVIDIA GB10), 128GBsealed appliance · 1 part · break down its parts	5 BLOCKERS	$5,378 (derived) DERIVEDMSI EdgeXpert (NVIDIA GB10), 128GB: landed costArithmetic on observed inputs only. No tunable assumption enters it. rig_landed = Σ part landed_total $5,378 = $5,378 MSI EdgeXpert (NVIDIA GB10), 128GB 537789 ¢ — SabrePC · ship Some part prices are older than 36 h. LOWER BOUND: 1 of 1 parts come from a retailer that does not state a delivery charge (SabrePC), so the rig costs at least this and never less.	$6,346 (estimate) ESTIMATEMSI EdgeXpert (NVIDIA GB10), 128GB: purchase plus five years of powerArithmetic that also depends on a tunable assumption, listed below. purchase_plus_5yr_power = purchase + five_year_power purchase_plus_5yr_power = $5,377.89 + $967.93 = $6,345.82 purchase_cents 537789 ¢ — rig_landed = Σ part landed_total five_year_power_cents 96793 ¢ — five_year_power = annual_kwh × ¢_per_kwh × 5 Some part prices are older than 36 h. LOWER BOUND: 1 of 1 parts come from a retailer that does not state a delivery charge (SabrePC), so the rig costs at least this and never less. Watts are at the wall (PSU conversion losses included by the caller). Load and idle draws are averages over their hours, not peaks. Duty cycle (Settings → Power): 6 h/day at load, 18 h/day idle. 365 days per year (leap days ignored, −0.07%). Flat rate for the whole period: no escalation, no time-of-use or tiered pricing, no demand charges. Rounded to the nearest cent, half-up, once from the unrounded annual cost.	$46.68 (derived) DERIVEDMSI EdgeXpert (NVIDIA GB10), 128GB: cost per GBArithmetic on observed inputs only. No tunable assumption enters it. cost_per_gb = landed_cost / usable_memory_gb = $5,378 / 115.2 GB = $46.68 /GB landed_cost 5377.89 USD — rig_landed = Σ part landed_total usable_memory_gb 115.2 GB — usable memory after the OS and framework reservation (Settings → Estimator) Some part prices are older than 36 h. LOWER BOUND: 1 of 1 parts come from a retailer that does not state a delivery charge (SabrePC), so the rig costs at least this and never less. Usable memory is below physical memory for at least one pool (OS/driver reservation; editable in Settings).	$19.70 (derived) DERIVEDMSI EdgeXpert (NVIDIA GB10), 128GB: cost per GB/sArithmetic on observed inputs only. No tunable assumption enters it. cost_per_gbs = landed_cost / memory_bandwidth_gbs = $5,378 / 273 GB/s = $19.7 per GB/s landed_cost 5377.89 USD — rig_landed = Σ part landed_total memory_bandwidth_gbs 273 GB/s — the published bandwidth of the rig's fastest memory pool Some part prices are older than 36 h. LOWER BOUND: 1 of 1 parts come from a retailer that does not state a delivery charge (SabrePC), so the rig costs at least this and never less. Weights are read from the fastest pool first; a rig whose weights spill into a slower pool decodes below this figure.	Unavailable — Cost per tok/s needs decode_tok_s: GLM-5.3 has no decode speed on this rig: Model needs 507.62 GB but MSI EdgeXpert (NVIDIA GB10), 128GB has 115.2 GB usable — it will not load, so there is no decode speed.
Priced fromSabrePCcondition newaccess tier 2, route sabrepc_ngstatestale 43 h — past the 36 hour freshness windowshipped
61	Lenovo ThinkStation PGX (NVIDIA GB10), 128GBsealed appliance · 1 part · broken down below	5 BLOCKERS	$5,118 (derived) DERIVEDLenovo ThinkStation PGX (NVIDIA GB10), 128GB: landed costArithmetic on observed inputs only. No tunable assumption enters it. rig_landed = Σ part landed_total $5,118 = $5,118 Lenovo ThinkStation PGX (NVIDIA GB10), 128GB 511799 ¢ — Central Computers · ship Some part prices are older than 36 h. LOWER BOUND: 1 of 1 parts come from a retailer that does not state a delivery charge (Central Computers), so the rig costs at least this and never less.	$6,086 (estimate) ESTIMATELenovo ThinkStation PGX (NVIDIA GB10), 128GB: purchase plus five years of powerArithmetic that also depends on a tunable assumption, listed below. purchase_plus_5yr_power = purchase + five_year_power purchase_plus_5yr_power = $5,117.99 + $967.93 = $6,085.92 purchase_cents 511799 ¢ — rig_landed = Σ part landed_total five_year_power_cents 96793 ¢ — five_year_power = annual_kwh × ¢_per_kwh × 5 Some part prices are older than 36 h. LOWER BOUND: 1 of 1 parts come from a retailer that does not state a delivery charge (Central Computers), so the rig costs at least this and never less. Watts are at the wall (PSU conversion losses included by the caller). Load and idle draws are averages over their hours, not peaks. Duty cycle (Settings → Power): 6 h/day at load, 18 h/day idle. 365 days per year (leap days ignored, −0.07%). Flat rate for the whole period: no escalation, no time-of-use or tiered pricing, no demand charges. Rounded to the nearest cent, half-up, once from the unrounded annual cost.	$44.43 (derived) DERIVEDLenovo ThinkStation PGX (NVIDIA GB10), 128GB: cost per GBArithmetic on observed inputs only. No tunable assumption enters it. cost_per_gb = landed_cost / usable_memory_gb = $5,118 / 115.2 GB = $44.43 /GB landed_cost 5117.99 USD — rig_landed = Σ part landed_total usable_memory_gb 115.2 GB — usable memory after the OS and framework reservation (Settings → Estimator) Some part prices are older than 36 h. LOWER BOUND: 1 of 1 parts come from a retailer that does not state a delivery charge (Central Computers), so the rig costs at least this and never less. Usable memory is below physical memory for at least one pool (OS/driver reservation; editable in Settings).	$18.75 (derived) DERIVEDLenovo ThinkStation PGX (NVIDIA GB10), 128GB: cost per GB/sArithmetic on observed inputs only. No tunable assumption enters it. cost_per_gbs = landed_cost / memory_bandwidth_gbs = $5,118 / 273 GB/s = $18.75 per GB/s landed_cost 5117.99 USD — rig_landed = Σ part landed_total memory_bandwidth_gbs 273 GB/s — the published bandwidth of the rig's fastest memory pool Some part prices are older than 36 h. LOWER BOUND: 1 of 1 parts come from a retailer that does not state a delivery charge (Central Computers), so the rig costs at least this and never less. Weights are read from the fastest pool first; a rig whose weights spill into a slower pool decodes below this figure.	Unavailable — Cost per tok/s needs decode_tok_s: GLM-5.3 has no decode speed on this rig: Model needs 507.62 GB but Lenovo ThinkStation PGX (NVIDIA GB10), 128GB has 115.2 GB usable — it will not load, so there is no decode speed.
Priced fromCentral Computerscondition newaccess tier 1, route jsonld_productstale 43 h — past the 36 hour freshness windowshipped
62	CPU · EPYC 9015 · 8× M321R8GA0BB0-CQK 64GB DDR5-4800 RDIMMDIY workstation · 6 parts · break down its parts	6 BLOCKERS	$5,037 (derived) DERIVEDCPU · EPYC 9015 · 8× M321R8GA0BB0-CQK 64GB DDR5-4800 RDIMM: landed costArithmetic on observed inputs only. No tunable assumption enters it. rig_landed = Σ part landed_total $638 + $895 + $2,534 + $75 + $192 + $704 = $5,037 AMD EPYC 9015 63809 ¢ — SabrePC · ship ASRock Rack GENOAD8QM3-2T/BCM 89458 ¢ — Newegg · ship Samsung M321R8GA0BB0-CQK 64GB DDR5-4800 RDIMM 253400 ¢ — SabrePC · ship ARCTIC Freezer 4U-SP5 7463 ¢ — ARCTIC · ship Phanteks Enthoo Pro 2 Server Edition 19191 ¢ — Newegg · ship Seasonic PRIME PX-2200 70371 ¢ — Newegg · ship Some part prices are older than 36 h. LOWER BOUND: 6 of 6 parts come from a retailer that does not state a delivery charge (SabrePC, Newegg, SabrePC, ARCTIC, Newegg, Newegg), so the rig costs at least this and never less.	Unavailable — Purchase + 5-year power needs the power term. Load draw at the wall is unavailable: Electrical feasibility cannot be computed exactly: no load or TDP watts for ASRock Rack GENOAD8QM3-2T/BCM, Samsung M321R8GA0BB0-CQK 64GB DDR5-4800 RDIMM, ARCTIC Freezer 4U-SP5, Phanteks Enthoo Pro 2 Server Edition. (missing: ASRock Rack GENOAD8QM3-2T/BCM: load_watts or tdp_watts, Samsung M321R8GA0BB0-CQK 64GB DDR5-4800 RDIMM: load_watts or tdp_watts, ARCTIC Freezer 4U-SP5: load_watts or tdp_watts, Phanteks Enthoo Pro 2 Server Edition: load_watts or tdp_watts, AMD EPYC 9015: idle_watts, ASRock Rack GENOAD8QM3-2T/BCM: idle_watts, Samsung M321R8GA0BB0-CQK 64GB DDR5-4800 RDIMM: idle_watts, ARCTIC Freezer 4U-SP5: idle_watts, Phanteks Enthoo Pro 2 Server Edition: idle_watts, Seasonic PRIME PX-2200: idle_watts)	$10.36 (derived) DERIVEDCPU · EPYC 9015 · 8× M321R8GA0BB0-CQK 64GB DDR5-4800 RDIMM: cost per GBArithmetic on observed inputs only. No tunable assumption enters it. cost_per_gb = landed_cost / usable_memory_gb = $5,037 / 486.4 GB = $10.36 /GB landed_cost 5036.92 USD — rig_landed = Σ part landed_total usable_memory_gb 486.4 GB — usable memory after the OS and framework reservation (Settings → Estimator) Some part prices are older than 36 h. LOWER BOUND: 6 of 6 parts come from a retailer that does not state a delivery charge (SabrePC, Newegg, SabrePC, ARCTIC, Newegg, Newegg), so the rig costs at least this and never less. Usable memory is below physical memory for at least one pool (OS/driver reservation; editable in Settings).	$16.40 (derived) DERIVEDCPU · EPYC 9015 · 8× M321R8GA0BB0-CQK 64GB DDR5-4800 RDIMM: cost per GB/sArithmetic on observed inputs only. No tunable assumption enters it. cost_per_gbs = landed_cost / memory_bandwidth_gbs = $5,037 / 307.2 GB/s = $16.4 per GB/s landed_cost 5036.92 USD — rig_landed = Σ part landed_total memory_bandwidth_gbs 307.2 GB/s — the published bandwidth of the rig's fastest memory pool Some part prices are older than 36 h. LOWER BOUND: 6 of 6 parts come from a retailer that does not state a delivery charge (SabrePC, Newegg, SabrePC, ARCTIC, Newegg, Newegg), so the rig costs at least this and never less. Weights are read from the fastest pool first; a rig whose weights spill into a slower pool decodes below this figure.	Unavailable — Cost per tok/s needs decode_tok_s: GLM-5.3 has no decode speed on this rig: Model needs 507.62 GB but DIY: AMD EPYC 9015 (CPU only) has 486.4 GB usable — it will not load, so there is no decode speed.
Priced from6 offers across 3 retailersoldeststale 44 h — past the 36 hour freshness windowopen the rig below for each part’s retailer, condition and route
63	CPU · EPYC 9015 · 8× M321R4GA3BB6-CQK 32GB DDR5-4800 RDIMMDIY workstation · 6 parts · break down its parts	6 BLOCKERS	$4,331 (derived) DERIVEDCPU · EPYC 9015 · 8× M321R4GA3BB6-CQK 32GB DDR5-4800 RDIMM: landed costArithmetic on observed inputs only. No tunable assumption enters it. rig_landed = Σ part landed_total $638 + $895 + $1,828 + $75 + $192 + $704 = $4,331 AMD EPYC 9015 63809 ¢ — SabrePC · ship ASRock Rack GENOAD8QM3-2T/BCM 89458 ¢ — Newegg · ship Samsung M321R4GA3BB6-CQK 32GB DDR5-4800 RDIMM 182776 ¢ — SabrePC · ship ARCTIC Freezer 4U-SP5 7463 ¢ — ARCTIC · ship Phanteks Enthoo Pro 2 Server Edition 19191 ¢ — Newegg · ship Seasonic PRIME PX-2200 70371 ¢ — Newegg · ship Some part prices are older than 36 h. LOWER BOUND: 6 of 6 parts come from a retailer that does not state a delivery charge (SabrePC, Newegg, SabrePC, ARCTIC, Newegg, Newegg), so the rig costs at least this and never less.	Unavailable — Purchase + 5-year power needs the power term. Load draw at the wall is unavailable: Electrical feasibility cannot be computed exactly: no load or TDP watts for ASRock Rack GENOAD8QM3-2T/BCM, Samsung M321R4GA3BB6-CQK 32GB DDR5-4800 RDIMM, ARCTIC Freezer 4U-SP5, Phanteks Enthoo Pro 2 Server Edition. (missing: ASRock Rack GENOAD8QM3-2T/BCM: load_watts or tdp_watts, Samsung M321R4GA3BB6-CQK 32GB DDR5-4800 RDIMM: load_watts or tdp_watts, ARCTIC Freezer 4U-SP5: load_watts or tdp_watts, Phanteks Enthoo Pro 2 Server Edition: load_watts or tdp_watts, AMD EPYC 9015: idle_watts, ASRock Rack GENOAD8QM3-2T/BCM: idle_watts, Samsung M321R4GA3BB6-CQK 32GB DDR5-4800 RDIMM: idle_watts, ARCTIC Freezer 4U-SP5: idle_watts, Phanteks Enthoo Pro 2 Server Edition: idle_watts, Seasonic PRIME PX-2200: idle_watts)	$17.81 (derived) DERIVEDCPU · EPYC 9015 · 8× M321R4GA3BB6-CQK 32GB DDR5-4800 RDIMM: cost per GBArithmetic on observed inputs only. No tunable assumption enters it. cost_per_gb = landed_cost / usable_memory_gb = $4,331 / 243.2 GB = $17.81 /GB landed_cost 4330.68 USD — rig_landed = Σ part landed_total usable_memory_gb 243.2 GB — usable memory after the OS and framework reservation (Settings → Estimator) Some part prices are older than 36 h. LOWER BOUND: 6 of 6 parts come from a retailer that does not state a delivery charge (SabrePC, Newegg, SabrePC, ARCTIC, Newegg, Newegg), so the rig costs at least this and never less. Usable memory is below physical memory for at least one pool (OS/driver reservation; editable in Settings).	$14.10 (derived) DERIVEDCPU · EPYC 9015 · 8× M321R4GA3BB6-CQK 32GB DDR5-4800 RDIMM: cost per GB/sArithmetic on observed inputs only. No tunable assumption enters it. cost_per_gbs = landed_cost / memory_bandwidth_gbs = $4,331 / 307.2 GB/s = $14.1 per GB/s landed_cost 4330.68 USD — rig_landed = Σ part landed_total memory_bandwidth_gbs 307.2 GB/s — the published bandwidth of the rig's fastest memory pool Some part prices are older than 36 h. LOWER BOUND: 6 of 6 parts come from a retailer that does not state a delivery charge (SabrePC, Newegg, SabrePC, ARCTIC, Newegg, Newegg), so the rig costs at least this and never less. Weights are read from the fastest pool first; a rig whose weights spill into a slower pool decodes below this figure.	Unavailable — Cost per tok/s needs decode_tok_s: GLM-5.3 has no decode speed on this rig: Model needs 507.62 GB but DIY: AMD EPYC 9015 (CPU only) has 243.2 GB usable — it will not load, so there is no decode speed.
Priced from6 offers across 3 retailersoldeststale 44 h — past the 36 hour freshness windowopen the rig below for each part’s retailer, condition and route
64	Minisforum MS-S1 MAX (Ryzen AI Max+ 395), 128GBsealed appliance · 1 part · break down its parts	6 BLOCKERS	$4,136 (derived) DERIVEDMinisforum MS-S1 MAX (Ryzen AI Max+ 395), 128GB: landed costArithmetic on observed inputs only. No tunable assumption enters it. rig_landed = Σ part landed_total $4,136 = $4,136 Minisforum MS-S1 MAX (Ryzen AI Max+ 395), 128GB 413598 ¢ — Minisforum · ship LOWER BOUND: 1 of 1 parts come from a retailer that does not state a delivery charge (Minisforum), so the rig costs at least this and never less.	Unavailable — Purchase + 5-year power needs the power term. Load draw at the wall is unavailable: Electrical feasibility cannot be computed exactly: no load or TDP watts for Minisforum MS-S1 MAX (Ryzen AI Max+ 395), 128GB. (missing: Minisforum MS-S1 MAX (Ryzen AI Max+ 395), 128GB: load_watts or tdp_watts, Minisforum MS-S1 MAX (Ryzen AI Max+ 395), 128GB: idle_watts)	$35.90 (derived) DERIVEDMinisforum MS-S1 MAX (Ryzen AI Max+ 395), 128GB: cost per GBArithmetic on observed inputs only. No tunable assumption enters it. cost_per_gb = landed_cost / usable_memory_gb = $4,136 / 115.2 GB = $35.9 /GB landed_cost 4135.98 USD — rig_landed = Σ part landed_total usable_memory_gb 115.2 GB — usable memory after the OS and framework reservation (Settings → Estimator) LOWER BOUND: 1 of 1 parts come from a retailer that does not state a delivery charge (Minisforum), so the rig costs at least this and never less. Usable memory is below physical memory for at least one pool (OS/driver reservation; editable in Settings).	$16.16 (derived) DERIVEDMinisforum MS-S1 MAX (Ryzen AI Max+ 395), 128GB: cost per GB/sArithmetic on observed inputs only. No tunable assumption enters it. cost_per_gbs = landed_cost / memory_bandwidth_gbs = $4,136 / 256 GB/s = $16.16 per GB/s landed_cost 4135.98 USD — rig_landed = Σ part landed_total memory_bandwidth_gbs 256 GB/s — the published bandwidth of the rig's fastest memory pool LOWER BOUND: 1 of 1 parts come from a retailer that does not state a delivery charge (Minisforum), so the rig costs at least this and never less. Weights are read from the fastest pool first; a rig whose weights spill into a slower pool decodes below this figure.	Unavailable — Cost per tok/s needs decode_tok_s: GLM-5.3 has no decode speed on this rig: Model needs 507.62 GB but Minisforum MS-S1 MAX (Ryzen AI Max+ 395), 128GB has 115.2 GB usable — it will not load, so there is no decode speed.
Priced fromMinisforumcondition newaccess tier 3, route shopify_store30 hshipped
65	GMKtec EVO-X3 (Ryzen AI Max+ 395), 128GBsealed appliance · 1 part · break down its parts	6 BLOCKERS	$4,052 (derived) DERIVEDGMKtec EVO-X3 (Ryzen AI Max+ 395), 128GB: landed costArithmetic on observed inputs only. No tunable assumption enters it. rig_landed = Σ part landed_total $4,052 = $4,052 GMKtec EVO-X3 (Ryzen AI Max+ 395), 128GB 405174 ¢ — GMKtec · ship LOWER BOUND: 1 of 1 parts come from a retailer that does not state a delivery charge (GMKtec), so the rig costs at least this and never less.	Unavailable — Purchase + 5-year power needs the power term. Load draw at the wall is unavailable: Electrical feasibility cannot be computed exactly: no load or TDP watts for GMKtec EVO-X3 (Ryzen AI Max+ 395), 128GB. (missing: GMKtec EVO-X3 (Ryzen AI Max+ 395), 128GB: load_watts or tdp_watts, GMKtec EVO-X3 (Ryzen AI Max+ 395), 128GB: idle_watts)	$35.17 (derived) DERIVEDGMKtec EVO-X3 (Ryzen AI Max+ 395), 128GB: cost per GBArithmetic on observed inputs only. No tunable assumption enters it. cost_per_gb = landed_cost / usable_memory_gb = $4,052 / 115.2 GB = $35.17 /GB landed_cost 4051.74 USD — rig_landed = Σ part landed_total usable_memory_gb 115.2 GB — usable memory after the OS and framework reservation (Settings → Estimator) LOWER BOUND: 1 of 1 parts come from a retailer that does not state a delivery charge (GMKtec), so the rig costs at least this and never less. Usable memory is below physical memory for at least one pool (OS/driver reservation; editable in Settings).	$15.83 (derived) DERIVEDGMKtec EVO-X3 (Ryzen AI Max+ 395), 128GB: cost per GB/sArithmetic on observed inputs only. No tunable assumption enters it. cost_per_gbs = landed_cost / memory_bandwidth_gbs = $4,052 / 256 GB/s = $15.83 per GB/s landed_cost 4051.74 USD — rig_landed = Σ part landed_total memory_bandwidth_gbs 256 GB/s — the published bandwidth of the rig's fastest memory pool LOWER BOUND: 1 of 1 parts come from a retailer that does not state a delivery charge (GMKtec), so the rig costs at least this and never less. Weights are read from the fastest pool first; a rig whose weights spill into a slower pool decodes below this figure.	Unavailable — Cost per tok/s needs decode_tok_s: GLM-5.3 has no decode speed on this rig: Model needs 507.62 GB but GMKtec EVO-X3 (Ryzen AI Max+ 395), 128GB has 115.2 GB usable — it will not load, so there is no decode speed.
Priced fromGMKteccondition newaccess tier 3, route shopify_store30 hshipped
66	Apple Mac Studio (2026), M5 Max 18-core CPU / 40-core GPU, 64GBsealed appliance · 1 part · break down its parts	5 BLOCKERS	$4,051 (derived) DERIVEDApple Mac Studio (2026), M5 Max 18-core CPU / 40-core GPU, 64GB: landed costArithmetic on observed inputs only. No tunable assumption enters it. rig_landed = Σ part landed_total $4,051 = $4,051 Apple Mac Studio (2026), M5 Max 18-core CPU / 40-core GPU, 64GB 405068 ¢ — Apple · pickup	$5,516 (estimate) ESTIMATEApple Mac Studio (2026), M5 Max 18-core CPU / 40-core GPU, 64GB: purchase plus five years of powerArithmetic that also depends on a tunable assumption, listed below. purchase_plus_5yr_power = purchase + five_year_power purchase_plus_5yr_power = $4,050.68 + $1,465.14 = $5,515.82 purchase_cents 405068 ¢ — rig_landed = Σ part landed_total five_year_power_cents 146514 ¢ — five_year_power = annual_kwh × ¢_per_kwh × 5 Watts are at the wall (PSU conversion losses included by the caller). Load and idle draws are averages over their hours, not peaks. Duty cycle (Settings → Power): 6 h/day at load, 18 h/day idle. 365 days per year (leap days ignored, −0.07%). Flat rate for the whole period: no escalation, no time-of-use or tiered pricing, no demand charges. Rounded to the nearest cent, half-up, once from the unrounded annual cost.	$84.39 (derived) DERIVEDApple Mac Studio (2026), M5 Max 18-core CPU / 40-core GPU, 64GB: cost per GBArithmetic on observed inputs only. No tunable assumption enters it. cost_per_gb = landed_cost / usable_memory_gb = $4,051 / 48 GB = $84.39 /GB landed_cost 4050.68 USD — rig_landed = Σ part landed_total usable_memory_gb 48 GB — usable memory after the OS and framework reservation (Settings → Estimator) Usable memory is below physical memory for at least one pool (OS/driver reservation; editable in Settings).	$6.60 (derived) DERIVEDApple Mac Studio (2026), M5 Max 18-core CPU / 40-core GPU, 64GB: cost per GB/sArithmetic on observed inputs only. No tunable assumption enters it. cost_per_gbs = landed_cost / memory_bandwidth_gbs = $4,051 / 614 GB/s = $6.6 per GB/s landed_cost 4050.68 USD — rig_landed = Σ part landed_total memory_bandwidth_gbs 614 GB/s — the published bandwidth of the rig's fastest memory pool Weights are read from the fastest pool first; a rig whose weights spill into a slower pool decodes below this figure.	Unavailable — Cost per tok/s needs decode_tok_s: GLM-5.3 has no decode speed on this rig: Model needs 507.62 GB but Apple Mac Studio (2026), M5 Max 18-core CPU / 40-core GPU, 64GB has 48 GB usable — it will not load, so there is no decode speed.
Priced fromApplecondition newaccess tier 3, route apple_store30 hlocal pickup beat shipping
67	Minisforum N5 MAX AI NAS (Ryzen AI Max+ 395), 128GBsealed appliance · 1 part · break down its parts	6 BLOCKERS	$3,837 (derived) DERIVEDMinisforum N5 MAX AI NAS (Ryzen AI Max+ 395), 128GB: landed costArithmetic on observed inputs only. No tunable assumption enters it. rig_landed = Σ part landed_total $3,837 = $3,837 Minisforum N5 MAX AI NAS (Ryzen AI Max+ 395), 128GB 383743 ¢ — Minisforum · ship LOWER BOUND: 1 of 1 parts come from a retailer that does not state a delivery charge (Minisforum), so the rig costs at least this and never less.	$4,482 (estimate) ESTIMATEMinisforum N5 MAX AI NAS (Ryzen AI Max+ 395), 128GB: purchase plus five years of powerArithmetic that also depends on a tunable assumption, listed below. purchase_plus_5yr_power = purchase + five_year_power purchase_plus_5yr_power = $3,837.43 + $644.31 = $4,481.74 purchase_cents 383743 ¢ — rig_landed = Σ part landed_total five_year_power_cents 64431 ¢ — five_year_power = annual_kwh × ¢_per_kwh × 5 LOWER BOUND: 1 of 1 parts come from a retailer that does not state a delivery charge (Minisforum), so the rig costs at least this and never less. Watts are at the wall (PSU conversion losses included by the caller). Load and idle draws are averages over their hours, not peaks. Duty cycle (Settings → Power): 6 h/day at load, 18 h/day idle. 365 days per year (leap days ignored, −0.07%). Flat rate for the whole period: no escalation, no time-of-use or tiered pricing, no demand charges. Rounded to the nearest cent, half-up, once from the unrounded annual cost.	$33.31 (derived) DERIVEDMinisforum N5 MAX AI NAS (Ryzen AI Max+ 395), 128GB: cost per GBArithmetic on observed inputs only. No tunable assumption enters it. cost_per_gb = landed_cost / usable_memory_gb = $3,837 / 115.2 GB = $33.31 /GB landed_cost 3837.43 USD — rig_landed = Σ part landed_total usable_memory_gb 115.2 GB — usable memory after the OS and framework reservation (Settings → Estimator) LOWER BOUND: 1 of 1 parts come from a retailer that does not state a delivery charge (Minisforum), so the rig costs at least this and never less. Usable memory is below physical memory for at least one pool (OS/driver reservation; editable in Settings).	$14.99 (derived) DERIVEDMinisforum N5 MAX AI NAS (Ryzen AI Max+ 395), 128GB: cost per GB/sArithmetic on observed inputs only. No tunable assumption enters it. cost_per_gbs = landed_cost / memory_bandwidth_gbs = $3,837 / 256 GB/s = $14.99 per GB/s landed_cost 3837.43 USD — rig_landed = Σ part landed_total memory_bandwidth_gbs 256 GB/s — the published bandwidth of the rig's fastest memory pool LOWER BOUND: 1 of 1 parts come from a retailer that does not state a delivery charge (Minisforum), so the rig costs at least this and never less. Weights are read from the fastest pool first; a rig whose weights spill into a slower pool decodes below this figure.	Unavailable — Cost per tok/s needs decode_tok_s: GLM-5.3 has no decode speed on this rig: Model needs 507.62 GB but Minisforum N5 MAX AI NAS (Ryzen AI Max+ 395), 128GB has 115.2 GB usable — it will not load, so there is no decode speed.
Priced fromMinisforumcondition newaccess tier 3, route shopify_store30 hshipped
68	GMKtec EVO-X2 (Ryzen AI Max+ 395), 128GBsealed appliance · 1 part · break down its parts	6 BLOCKERS	$3,732 (derived) DERIVEDGMKtec EVO-X2 (Ryzen AI Max+ 395), 128GB: landed costArithmetic on observed inputs only. No tunable assumption enters it. rig_landed = Σ part landed_total $3,732 = $3,732 GMKtec EVO-X2 (Ryzen AI Max+ 395), 128GB 373186 ¢ — GMKtec · ship LOWER BOUND: 1 of 1 parts come from a retailer that does not state a delivery charge (GMKtec), so the rig costs at least this and never less.	Unavailable — Purchase + 5-year power needs the power term. Idle draw is not published for 1 of 1 part in this rig, and the duty cycle in Settings spends most of the day idle — counting an unknown idle draw as zero would understate the bill. (missing: GMKtec EVO-X2 (Ryzen AI Max+ 395), 128GB: idle_watts)	$32.39 (derived) DERIVEDGMKtec EVO-X2 (Ryzen AI Max+ 395), 128GB: cost per GBArithmetic on observed inputs only. No tunable assumption enters it. cost_per_gb = landed_cost / usable_memory_gb = $3,732 / 115.2 GB = $32.39 /GB landed_cost 3731.86 USD — rig_landed = Σ part landed_total usable_memory_gb 115.2 GB — usable memory after the OS and framework reservation (Settings → Estimator) LOWER BOUND: 1 of 1 parts come from a retailer that does not state a delivery charge (GMKtec), so the rig costs at least this and never less. Usable memory is below physical memory for at least one pool (OS/driver reservation; editable in Settings).	$14.58 (derived) DERIVEDGMKtec EVO-X2 (Ryzen AI Max+ 395), 128GB: cost per GB/sArithmetic on observed inputs only. No tunable assumption enters it. cost_per_gbs = landed_cost / memory_bandwidth_gbs = $3,732 / 256 GB/s = $14.58 per GB/s landed_cost 3731.86 USD — rig_landed = Σ part landed_total memory_bandwidth_gbs 256 GB/s — the published bandwidth of the rig's fastest memory pool LOWER BOUND: 1 of 1 parts come from a retailer that does not state a delivery charge (GMKtec), so the rig costs at least this and never less. Weights are read from the fastest pool first; a rig whose weights spill into a slower pool decodes below this figure.	Unavailable — Cost per tok/s needs decode_tok_s: GLM-5.3 has no decode speed on this rig: Model needs 507.62 GB but GMKtec EVO-X2 (Ryzen AI Max+ 395), 128GB has 115.2 GB usable — it will not load, so there is no decode speed.
Priced fromGMKteccondition newaccess tier 3, route shopify_store30 hshipped
69	Framework Framework Desktop DIY Edition (Ryzen AI Max+ 395), 128GBsealed appliance · 1 part · break down its parts	6 BLOCKERS	$3,678 (derived) DERIVEDFramework Framework Desktop DIY Edition (Ryzen AI Max+ 395), 128GB: landed costArithmetic on observed inputs only. No tunable assumption enters it. rig_landed = Σ part landed_total $3,678 = $3,678 Framework Framework Desktop DIY Edition (Ryzen AI Max+ 395), 128GB 367750 ¢ — Framework · ship Some part prices are older than 36 h. LOWER BOUND: 1 of 1 parts come from a retailer that does not state a delivery charge (Framework), so the rig costs at least this and never less.	Unavailable — Purchase + 5-year power needs the power term. Idle draw is not published for 1 of 1 part in this rig, and the duty cycle in Settings spends most of the day idle — counting an unknown idle draw as zero would understate the bill. (missing: Framework Framework Desktop DIY Edition (Ryzen AI Max+ 395), 128GB: idle_watts)	$31.92 (derived) DERIVEDFramework Framework Desktop DIY Edition (Ryzen AI Max+ 395), 128GB: cost per GBArithmetic on observed inputs only. No tunable assumption enters it. cost_per_gb = landed_cost / usable_memory_gb = $3,678 / 115.2 GB = $31.92 /GB landed_cost 3677.5 USD — rig_landed = Σ part landed_total usable_memory_gb 115.2 GB — usable memory after the OS and framework reservation (Settings → Estimator) Some part prices are older than 36 h. LOWER BOUND: 1 of 1 parts come from a retailer that does not state a delivery charge (Framework), so the rig costs at least this and never less. Usable memory is below physical memory for at least one pool (OS/driver reservation; editable in Settings).	$14.37 (derived) DERIVEDFramework Framework Desktop DIY Edition (Ryzen AI Max+ 395), 128GB: cost per GB/sArithmetic on observed inputs only. No tunable assumption enters it. cost_per_gbs = landed_cost / memory_bandwidth_gbs = $3,678 / 256 GB/s = $14.37 per GB/s landed_cost 3677.5 USD — rig_landed = Σ part landed_total memory_bandwidth_gbs 256 GB/s — the published bandwidth of the rig's fastest memory pool Some part prices are older than 36 h. LOWER BOUND: 1 of 1 parts come from a retailer that does not state a delivery charge (Framework), so the rig costs at least this and never less. Weights are read from the fastest pool first; a rig whose weights spill into a slower pool decodes below this figure.	Unavailable — Cost per tok/s needs decode_tok_s: GLM-5.3 has no decode speed on this rig: Model needs 507.62 GB but Framework Framework Desktop DIY Edition (Ryzen AI Max+ 395), 128GB has 115.2 GB usable — it will not load, so there is no decode speed.
Priced fromFrameworkcondition newaccess tier 2, route framework_htmlstale 44 h — past the 36 hour freshness windowshipped
70	Apple Mac mini (2026), M5 Pro 18-core CPU / 20-core GPU, 64GBsealed appliance · 1 part · break down its parts	5 BLOCKERS	$3,411 (derived) DERIVEDApple Mac mini (2026), M5 Pro 18-core CPU / 20-core GPU, 64GB: landed costArithmetic on observed inputs only. No tunable assumption enters it. rig_landed = Σ part landed_total $3,411 = $3,411 Apple Mac mini (2026), M5 Pro 18-core CPU / 20-core GPU, 64GB 341093 ¢ — Apple · pickup	$3,911 (estimate) ESTIMATEApple Mac mini (2026), M5 Pro 18-core CPU / 20-core GPU, 64GB: purchase plus five years of powerArithmetic that also depends on a tunable assumption, listed below. purchase_plus_5yr_power = purchase + five_year_power purchase_plus_5yr_power = $3,410.93 + $500.15 = $3,911.08 purchase_cents 341093 ¢ — rig_landed = Σ part landed_total five_year_power_cents 50015 ¢ — five_year_power = annual_kwh × ¢_per_kwh × 5 Watts are at the wall (PSU conversion losses included by the caller). Load and idle draws are averages over their hours, not peaks. Duty cycle (Settings → Power): 6 h/day at load, 18 h/day idle. 365 days per year (leap days ignored, −0.07%). Flat rate for the whole period: no escalation, no time-of-use or tiered pricing, no demand charges. Rounded to the nearest cent, half-up, once from the unrounded annual cost.	$71.06 (derived) DERIVEDApple Mac mini (2026), M5 Pro 18-core CPU / 20-core GPU, 64GB: cost per GBArithmetic on observed inputs only. No tunable assumption enters it. cost_per_gb = landed_cost / usable_memory_gb = $3,411 / 48 GB = $71.06 /GB landed_cost 3410.93 USD — rig_landed = Σ part landed_total usable_memory_gb 48 GB — usable memory after the OS and framework reservation (Settings → Estimator) Usable memory is below physical memory for at least one pool (OS/driver reservation; editable in Settings).	$11.11 (derived) DERIVEDApple Mac mini (2026), M5 Pro 18-core CPU / 20-core GPU, 64GB: cost per GB/sArithmetic on observed inputs only. No tunable assumption enters it. cost_per_gbs = landed_cost / memory_bandwidth_gbs = $3,411 / 307 GB/s = $11.11 per GB/s landed_cost 3410.93 USD — rig_landed = Σ part landed_total memory_bandwidth_gbs 307 GB/s — the published bandwidth of the rig's fastest memory pool Weights are read from the fastest pool first; a rig whose weights spill into a slower pool decodes below this figure.	Unavailable — Cost per tok/s needs decode_tok_s: GLM-5.3 has no decode speed on this rig: Model needs 507.62 GB but Apple Mac mini (2026), M5 Pro 18-core CPU / 20-core GPU, 64GB has 48 GB usable — it will not load, so there is no decode speed.
Priced fromApplecondition newaccess tier 3, route apple_store30 hlocal pickup beat shipping
71	Apple Mac Studio (2026), M5 Max 18-core CPU / 40-core GPU, 48GBsealed appliance · 1 part · break down its parts	5 BLOCKERS	$3,304 (derived) DERIVEDApple Mac Studio (2026), M5 Max 18-core CPU / 40-core GPU, 48GB: landed costArithmetic on observed inputs only. No tunable assumption enters it. rig_landed = Σ part landed_total $3,304 = $3,304 Apple Mac Studio (2026), M5 Max 18-core CPU / 40-core GPU, 48GB 330431 ¢ — Apple · ship LOWER BOUND: 1 of 1 parts come from a retailer that does not state a delivery charge (Apple), so the rig costs at least this and never less.	$4,769 (estimate) ESTIMATEApple Mac Studio (2026), M5 Max 18-core CPU / 40-core GPU, 48GB: purchase plus five years of powerArithmetic that also depends on a tunable assumption, listed below. purchase_plus_5yr_power = purchase + five_year_power purchase_plus_5yr_power = $3,304.31 + $1,465.14 = $4,769.45 purchase_cents 330431 ¢ — rig_landed = Σ part landed_total five_year_power_cents 146514 ¢ — five_year_power = annual_kwh × ¢_per_kwh × 5 LOWER BOUND: 1 of 1 parts come from a retailer that does not state a delivery charge (Apple), so the rig costs at least this and never less. Watts are at the wall (PSU conversion losses included by the caller). Load and idle draws are averages over their hours, not peaks. Duty cycle (Settings → Power): 6 h/day at load, 18 h/day idle. 365 days per year (leap days ignored, −0.07%). Flat rate for the whole period: no escalation, no time-of-use or tiered pricing, no demand charges. Rounded to the nearest cent, half-up, once from the unrounded annual cost.	$91.79 (derived) DERIVEDApple Mac Studio (2026), M5 Max 18-core CPU / 40-core GPU, 48GB: cost per GBArithmetic on observed inputs only. No tunable assumption enters it. cost_per_gb = landed_cost / usable_memory_gb = $3,304 / 36 GB = $91.79 /GB landed_cost 3304.31 USD — rig_landed = Σ part landed_total usable_memory_gb 36 GB — usable memory after the OS and framework reservation (Settings → Estimator) LOWER BOUND: 1 of 1 parts come from a retailer that does not state a delivery charge (Apple), so the rig costs at least this and never less. Usable memory is below physical memory for at least one pool (OS/driver reservation; editable in Settings).	$5.38 (derived) DERIVEDApple Mac Studio (2026), M5 Max 18-core CPU / 40-core GPU, 48GB: cost per GB/sArithmetic on observed inputs only. No tunable assumption enters it. cost_per_gbs = landed_cost / memory_bandwidth_gbs = $3,304 / 614 GB/s = $5.38 per GB/s landed_cost 3304.31 USD — rig_landed = Σ part landed_total memory_bandwidth_gbs 614 GB/s — the published bandwidth of the rig's fastest memory pool LOWER BOUND: 1 of 1 parts come from a retailer that does not state a delivery charge (Apple), so the rig costs at least this and never less. Weights are read from the fastest pool first; a rig whose weights spill into a slower pool decodes below this figure.	Unavailable — Cost per tok/s needs decode_tok_s: GLM-5.3 has no decode speed on this rig: Model needs 507.62 GB but Apple Mac Studio (2026), M5 Max 18-core CPU / 40-core GPU, 48GB has 36 GB usable — it will not load, so there is no decode speed.
Priced fromApplecondition newaccess tier 3, route apple_store30 hshipped
72	Apple Mac mini (2026), M5 Pro 15-core CPU / 16-core GPU, 64GBsealed appliance · 1 part · break down its parts	5 BLOCKERS	$2,878 (derived) DERIVEDApple Mac mini (2026), M5 Pro 15-core CPU / 16-core GPU, 64GB: landed costArithmetic on observed inputs only. No tunable assumption enters it. rig_landed = Σ part landed_total $2,878 = $2,878 Apple Mac mini (2026), M5 Pro 15-core CPU / 16-core GPU, 64GB 287781 ¢ — Apple · ship LOWER BOUND: 1 of 1 parts come from a retailer that does not state a delivery charge (Apple), so the rig costs at least this and never less.	Unavailable — Purchase + 5-year power needs the power term. Idle draw is not published for 1 of 1 part in this rig, and the duty cycle in Settings spends most of the day idle — counting an unknown idle draw as zero would understate the bill. (missing: Apple Mac mini (2026), M5 Pro 15-core CPU / 16-core GPU, 64GB: idle_watts)	$59.95 (derived) DERIVEDApple Mac mini (2026), M5 Pro 15-core CPU / 16-core GPU, 64GB: cost per GBArithmetic on observed inputs only. No tunable assumption enters it. cost_per_gb = landed_cost / usable_memory_gb = $2,878 / 48 GB = $59.95 /GB landed_cost 2877.81 USD — rig_landed = Σ part landed_total usable_memory_gb 48 GB — usable memory after the OS and framework reservation (Settings → Estimator) LOWER BOUND: 1 of 1 parts come from a retailer that does not state a delivery charge (Apple), so the rig costs at least this and never less. Usable memory is below physical memory for at least one pool (OS/driver reservation; editable in Settings).	$9.37 (derived) DERIVEDApple Mac mini (2026), M5 Pro 15-core CPU / 16-core GPU, 64GB: cost per GB/sArithmetic on observed inputs only. No tunable assumption enters it. cost_per_gbs = landed_cost / memory_bandwidth_gbs = $2,878 / 307 GB/s = $9.37 per GB/s landed_cost 2877.81 USD — rig_landed = Σ part landed_total memory_bandwidth_gbs 307 GB/s — the published bandwidth of the rig's fastest memory pool LOWER BOUND: 1 of 1 parts come from a retailer that does not state a delivery charge (Apple), so the rig costs at least this and never less. Weights are read from the fastest pool first; a rig whose weights spill into a slower pool decodes below this figure.	Unavailable — Cost per tok/s needs decode_tok_s: GLM-5.3 has no decode speed on this rig: Model needs 507.62 GB but Apple Mac mini (2026), M5 Pro 15-core CPU / 16-core GPU, 64GB has 48 GB usable — it will not load, so there is no decode speed.
Priced fromApplecondition newaccess tier 3, route apple_store30 hshipped
73	Minisforum MS-S1 MAX (Ryzen AI Max+ 395), 64GBsealed appliance · 1 part · break down its parts	6 BLOCKERS	$2,771 (derived) DERIVEDMinisforum MS-S1 MAX (Ryzen AI Max+ 395), 64GB: landed costArithmetic on observed inputs only. No tunable assumption enters it. rig_landed = Σ part landed_total $2,771 = $2,771 Minisforum MS-S1 MAX (Ryzen AI Max+ 395), 64GB 277118 ¢ — Minisforum · ship LOWER BOUND: 1 of 1 parts come from a retailer that does not state a delivery charge (Minisforum), so the rig costs at least this and never less.	Unavailable — Purchase + 5-year power needs the power term. Load draw at the wall is unavailable: Electrical feasibility cannot be computed exactly: no load or TDP watts for Minisforum MS-S1 MAX (Ryzen AI Max+ 395), 64GB. (missing: Minisforum MS-S1 MAX (Ryzen AI Max+ 395), 64GB: load_watts or tdp_watts, Minisforum MS-S1 MAX (Ryzen AI Max+ 395), 64GB: idle_watts)	$48.11 (derived) DERIVEDMinisforum MS-S1 MAX (Ryzen AI Max+ 395), 64GB: cost per GBArithmetic on observed inputs only. No tunable assumption enters it. cost_per_gb = landed_cost / usable_memory_gb = $2,771 / 57.6 GB = $48.11 /GB landed_cost 2771.18 USD — rig_landed = Σ part landed_total usable_memory_gb 57.6 GB — usable memory after the OS and framework reservation (Settings → Estimator) LOWER BOUND: 1 of 1 parts come from a retailer that does not state a delivery charge (Minisforum), so the rig costs at least this and never less. Usable memory is below physical memory for at least one pool (OS/driver reservation; editable in Settings).	$10.82 (derived) DERIVEDMinisforum MS-S1 MAX (Ryzen AI Max+ 395), 64GB: cost per GB/sArithmetic on observed inputs only. No tunable assumption enters it. cost_per_gbs = landed_cost / memory_bandwidth_gbs = $2,771 / 256 GB/s = $10.82 per GB/s landed_cost 2771.18 USD — rig_landed = Σ part landed_total memory_bandwidth_gbs 256 GB/s — the published bandwidth of the rig's fastest memory pool LOWER BOUND: 1 of 1 parts come from a retailer that does not state a delivery charge (Minisforum), so the rig costs at least this and never less. Weights are read from the fastest pool first; a rig whose weights spill into a slower pool decodes below this figure.	Unavailable — Cost per tok/s needs decode_tok_s: GLM-5.3 has no decode speed on this rig: Model needs 507.62 GB but Minisforum MS-S1 MAX (Ryzen AI Max+ 395), 64GB has 57.6 GB usable — it will not load, so there is no decode speed.
Priced fromMinisforumcondition newaccess tier 3, route shopify_store30 hshipped
74	Apple Mac Studio (2026), M5 Max 18-core CPU / 32-core GPU, 36GBsealed appliance · 1 part · break down its parts	6 BLOCKERS	$2,665 (derived) DERIVEDApple Mac Studio (2026), M5 Max 18-core CPU / 32-core GPU, 36GB: landed costArithmetic on observed inputs only. No tunable assumption enters it. rig_landed = Σ part landed_total $2,665 = $2,665 Apple Mac Studio (2026), M5 Max 18-core CPU / 32-core GPU, 36GB 266456 ¢ — Apple · pickup	$3,306 (estimate) ESTIMATEApple Mac Studio (2026), M5 Max 18-core CPU / 32-core GPU, 36GB: purchase plus five years of powerArithmetic that also depends on a tunable assumption, listed below. purchase_plus_5yr_power = purchase + five_year_power purchase_plus_5yr_power = $2,664.56 + $641.37 = $3,305.93 purchase_cents 266456 ¢ — rig_landed = Σ part landed_total five_year_power_cents 64137 ¢ — five_year_power = annual_kwh × ¢_per_kwh × 5 Watts are at the wall (PSU conversion losses included by the caller). Load and idle draws are averages over their hours, not peaks. Duty cycle (Settings → Power): 6 h/day at load, 18 h/day idle. 365 days per year (leap days ignored, −0.07%). Flat rate for the whole period: no escalation, no time-of-use or tiered pricing, no demand charges. Rounded to the nearest cent, half-up, once from the unrounded annual cost.	$98.69 (derived) DERIVEDApple Mac Studio (2026), M5 Max 18-core CPU / 32-core GPU, 36GB: cost per GBArithmetic on observed inputs only. No tunable assumption enters it. cost_per_gb = landed_cost / usable_memory_gb = $2,665 / 27 GB = $98.69 /GB landed_cost 2664.56 USD — rig_landed = Σ part landed_total usable_memory_gb 27 GB — usable memory after the OS and framework reservation (Settings → Estimator) Usable memory is below physical memory for at least one pool (OS/driver reservation; editable in Settings).	$5.79 (derived) DERIVEDApple Mac Studio (2026), M5 Max 18-core CPU / 32-core GPU, 36GB: cost per GB/sArithmetic on observed inputs only. No tunable assumption enters it. cost_per_gbs = landed_cost / memory_bandwidth_gbs = $2,665 / 460 GB/s = $5.79 per GB/s landed_cost 2664.56 USD — rig_landed = Σ part landed_total memory_bandwidth_gbs 460 GB/s — the published bandwidth of the rig's fastest memory pool Weights are read from the fastest pool first; a rig whose weights spill into a slower pool decodes below this figure.	Unavailable — Cost per tok/s needs decode_tok_s: GLM-5.3 has no decode speed on this rig: Model needs 507.62 GB but Apple Mac Studio (2026), M5 Max 18-core CPU / 32-core GPU, 36GB has 27 GB usable — it will not load, so there is no decode speed.
Priced fromApplecondition newaccess tier 3, route apple_store30 hlocal pickup beat shipping
75	Minisforum N5 MAX AI NAS (Ryzen AI Max+ 395), 64GBsealed appliance · 1 part · break down its parts	6 BLOCKERS	$2,558 (derived) DERIVEDMinisforum N5 MAX AI NAS (Ryzen AI Max+ 395), 64GB: landed costArithmetic on observed inputs only. No tunable assumption enters it. rig_landed = Σ part landed_total $2,558 = $2,558 Minisforum N5 MAX AI NAS (Ryzen AI Max+ 395), 64GB 255793 ¢ — Minisforum · ship LOWER BOUND: 1 of 1 parts come from a retailer that does not state a delivery charge (Minisforum), so the rig costs at least this and never less.	$3,202 (estimate) ESTIMATEMinisforum N5 MAX AI NAS (Ryzen AI Max+ 395), 64GB: purchase plus five years of powerArithmetic that also depends on a tunable assumption, listed below. purchase_plus_5yr_power = purchase + five_year_power purchase_plus_5yr_power = $2,557.93 + $644.31 = $3,202.24 purchase_cents 255793 ¢ — rig_landed = Σ part landed_total five_year_power_cents 64431 ¢ — five_year_power = annual_kwh × ¢_per_kwh × 5 LOWER BOUND: 1 of 1 parts come from a retailer that does not state a delivery charge (Minisforum), so the rig costs at least this and never less. Watts are at the wall (PSU conversion losses included by the caller). Load and idle draws are averages over their hours, not peaks. Duty cycle (Settings → Power): 6 h/day at load, 18 h/day idle. 365 days per year (leap days ignored, −0.07%). Flat rate for the whole period: no escalation, no time-of-use or tiered pricing, no demand charges. Rounded to the nearest cent, half-up, once from the unrounded annual cost.	$44.41 (derived) DERIVEDMinisforum N5 MAX AI NAS (Ryzen AI Max+ 395), 64GB: cost per GBArithmetic on observed inputs only. No tunable assumption enters it. cost_per_gb = landed_cost / usable_memory_gb = $2,558 / 57.6 GB = $44.41 /GB landed_cost 2557.93 USD — rig_landed = Σ part landed_total usable_memory_gb 57.6 GB — usable memory after the OS and framework reservation (Settings → Estimator) LOWER BOUND: 1 of 1 parts come from a retailer that does not state a delivery charge (Minisforum), so the rig costs at least this and never less. Usable memory is below physical memory for at least one pool (OS/driver reservation; editable in Settings).	$9.99 (derived) DERIVEDMinisforum N5 MAX AI NAS (Ryzen AI Max+ 395), 64GB: cost per GB/sArithmetic on observed inputs only. No tunable assumption enters it. cost_per_gbs = landed_cost / memory_bandwidth_gbs = $2,558 / 256 GB/s = $9.99 per GB/s landed_cost 2557.93 USD — rig_landed = Σ part landed_total memory_bandwidth_gbs 256 GB/s — the published bandwidth of the rig's fastest memory pool LOWER BOUND: 1 of 1 parts come from a retailer that does not state a delivery charge (Minisforum), so the rig costs at least this and never less. Weights are read from the fastest pool first; a rig whose weights spill into a slower pool decodes below this figure.	Unavailable — Cost per tok/s needs decode_tok_s: GLM-5.3 has no decode speed on this rig: Model needs 507.62 GB but Minisforum N5 MAX AI NAS (Ryzen AI Max+ 395), 64GB has 57.6 GB usable — it will not load, so there is no decode speed.
Priced fromMinisforumcondition newaccess tier 3, route shopify_store30 hshipped
76	GMKtec EVO-X2 (Ryzen AI Max+ 395), 64GBsealed appliance · 1 part · break down its parts	6 BLOCKERS	$2,346 (derived) DERIVEDGMKtec EVO-X2 (Ryzen AI Max+ 395), 64GB: landed costArithmetic on observed inputs only. No tunable assumption enters it. rig_landed = Σ part landed_total $2,346 = $2,346 GMKtec EVO-X2 (Ryzen AI Max+ 395), 64GB 234574 ¢ — GMKtec · ship LOWER BOUND: 1 of 1 parts come from a retailer that does not state a delivery charge (GMKtec), so the rig costs at least this and never less.	Unavailable — Purchase + 5-year power needs the power term. Idle draw is not published for 1 of 1 part in this rig, and the duty cycle in Settings spends most of the day idle — counting an unknown idle draw as zero would understate the bill. (missing: GMKtec EVO-X2 (Ryzen AI Max+ 395), 64GB: idle_watts)	$40.72 (derived) DERIVEDGMKtec EVO-X2 (Ryzen AI Max+ 395), 64GB: cost per GBArithmetic on observed inputs only. No tunable assumption enters it. cost_per_gb = landed_cost / usable_memory_gb = $2,346 / 57.6 GB = $40.72 /GB landed_cost 2345.74 USD — rig_landed = Σ part landed_total usable_memory_gb 57.6 GB — usable memory after the OS and framework reservation (Settings → Estimator) LOWER BOUND: 1 of 1 parts come from a retailer that does not state a delivery charge (GMKtec), so the rig costs at least this and never less. Usable memory is below physical memory for at least one pool (OS/driver reservation; editable in Settings).	$9.16 (derived) DERIVEDGMKtec EVO-X2 (Ryzen AI Max+ 395), 64GB: cost per GB/sArithmetic on observed inputs only. No tunable assumption enters it. cost_per_gbs = landed_cost / memory_bandwidth_gbs = $2,346 / 256 GB/s = $9.16 per GB/s landed_cost 2345.74 USD — rig_landed = Σ part landed_total memory_bandwidth_gbs 256 GB/s — the published bandwidth of the rig's fastest memory pool LOWER BOUND: 1 of 1 parts come from a retailer that does not state a delivery charge (GMKtec), so the rig costs at least this and never less. Weights are read from the fastest pool first; a rig whose weights spill into a slower pool decodes below this figure.	Unavailable — Cost per tok/s needs decode_tok_s: GLM-5.3 has no decode speed on this rig: Model needs 507.62 GB but GMKtec EVO-X2 (Ryzen AI Max+ 395), 64GB has 57.6 GB usable — it will not load, so there is no decode speed.
Priced fromGMKteccondition newaccess tier 3, route shopify_store30 hshipped
77	Framework Framework Desktop DIY Edition (Ryzen AI Max+ 395), 64GBsealed appliance · 1 part · break down its parts	6 BLOCKERS	$2,089 (derived) DERIVEDFramework Framework Desktop DIY Edition (Ryzen AI Max+ 395), 64GB: landed costArithmetic on observed inputs only. No tunable assumption enters it. rig_landed = Σ part landed_total $2,089 = $2,089 Framework Framework Desktop DIY Edition (Ryzen AI Max+ 395), 64GB 208878 ¢ — Framework · ship Some part prices are older than 36 h. LOWER BOUND: 1 of 1 parts come from a retailer that does not state a delivery charge (Framework), so the rig costs at least this and never less.	Unavailable — Purchase + 5-year power needs the power term. Idle draw is not published for 1 of 1 part in this rig, and the duty cycle in Settings spends most of the day idle — counting an unknown idle draw as zero would understate the bill. (missing: Framework Framework Desktop DIY Edition (Ryzen AI Max+ 395), 64GB: idle_watts)	$36.26 (derived) DERIVEDFramework Framework Desktop DIY Edition (Ryzen AI Max+ 395), 64GB: cost per GBArithmetic on observed inputs only. No tunable assumption enters it. cost_per_gb = landed_cost / usable_memory_gb = $2,089 / 57.6 GB = $36.26 /GB landed_cost 2088.78 USD — rig_landed = Σ part landed_total usable_memory_gb 57.6 GB — usable memory after the OS and framework reservation (Settings → Estimator) Some part prices are older than 36 h. LOWER BOUND: 1 of 1 parts come from a retailer that does not state a delivery charge (Framework), so the rig costs at least this and never less. Usable memory is below physical memory for at least one pool (OS/driver reservation; editable in Settings).	$8.16 (derived) DERIVEDFramework Framework Desktop DIY Edition (Ryzen AI Max+ 395), 64GB: cost per GB/sArithmetic on observed inputs only. No tunable assumption enters it. cost_per_gbs = landed_cost / memory_bandwidth_gbs = $2,089 / 256 GB/s = $8.16 per GB/s landed_cost 2088.78 USD — rig_landed = Σ part landed_total memory_bandwidth_gbs 256 GB/s — the published bandwidth of the rig's fastest memory pool Some part prices are older than 36 h. LOWER BOUND: 1 of 1 parts come from a retailer that does not state a delivery charge (Framework), so the rig costs at least this and never less. Weights are read from the fastest pool first; a rig whose weights spill into a slower pool decodes below this figure.	Unavailable — Cost per tok/s needs decode_tok_s: GLM-5.3 has no decode speed on this rig: Model needs 507.62 GB but Framework Framework Desktop DIY Edition (Ryzen AI Max+ 395), 64GB has 57.6 GB usable — it will not load, so there is no decode speed.
Priced fromFrameworkcondition newaccess tier 2, route framework_htmlstale 44 h — past the 36 hour freshness windowshipped
78	Apple Mac mini (2026), M6, 32GBsealed appliance · 1 part · break down its parts	6 BLOCKERS	$1,918 (derived) DERIVEDApple Mac mini (2026), M6, 32GB: landed costArithmetic on observed inputs only. No tunable assumption enters it. rig_landed = Σ part landed_total $1,918 = $1,918 Apple Mac mini (2026), M6, 32GB 191818 ¢ — Apple · pickup	$2,410 (estimate) ESTIMATEApple Mac mini (2026), M6, 32GB: purchase plus five years of powerArithmetic that also depends on a tunable assumption, listed below. purchase_plus_5yr_power = purchase + five_year_power purchase_plus_5yr_power = $1,918.18 + $491.32 = $2,409.50 purchase_cents 191818 ¢ — rig_landed = Σ part landed_total five_year_power_cents 49132 ¢ — five_year_power = annual_kwh × ¢_per_kwh × 5 Watts are at the wall (PSU conversion losses included by the caller). Load and idle draws are averages over their hours, not peaks. Duty cycle (Settings → Power): 6 h/day at load, 18 h/day idle. 365 days per year (leap days ignored, −0.07%). Flat rate for the whole period: no escalation, no time-of-use or tiered pricing, no demand charges. Rounded to the nearest cent, half-up, once from the unrounded annual cost.	$79.92 (derived) DERIVEDApple Mac mini (2026), M6, 32GB: cost per GBArithmetic on observed inputs only. No tunable assumption enters it. cost_per_gb = landed_cost / usable_memory_gb = $1,918 / 24 GB = $79.92 /GB landed_cost 1918.18 USD — rig_landed = Σ part landed_total usable_memory_gb 24 GB — usable memory after the OS and framework reservation (Settings → Estimator) Usable memory is below physical memory for at least one pool (OS/driver reservation; editable in Settings).	$11.28 (derived) DERIVEDApple Mac mini (2026), M6, 32GB: cost per GB/sArithmetic on observed inputs only. No tunable assumption enters it. cost_per_gbs = landed_cost / memory_bandwidth_gbs = $1,918 / 170 GB/s = $11.28 per GB/s landed_cost 1918.18 USD — rig_landed = Σ part landed_total memory_bandwidth_gbs 170 GB/s — the published bandwidth of the rig's fastest memory pool Weights are read from the fastest pool first; a rig whose weights spill into a slower pool decodes below this figure.	Unavailable — Cost per tok/s needs decode_tok_s: GLM-5.3 has no decode speed on this rig: Model needs 507.62 GB but Apple Mac mini (2026), M6, 32GB has 24 GB usable — it will not load, so there is no decode speed.
Priced fromApplecondition newaccess tier 3, route apple_store30 hlocal pickup beat shipping
79	Apple Mac mini (2026), M5 Pro 15-core CPU / 16-core GPU, 24GBsealed appliance · 1 part · break down its parts	6 BLOCKERS	$1,812 (derived) DERIVEDApple Mac mini (2026), M5 Pro 15-core CPU / 16-core GPU, 24GB: landed costArithmetic on observed inputs only. No tunable assumption enters it. rig_landed = Σ part landed_total $1,812 = $1,812 Apple Mac mini (2026), M5 Pro 15-core CPU / 16-core GPU, 24GB 181156 ¢ — Apple · pickup	Unavailable — Purchase + 5-year power needs the power term. Idle draw is not published for 1 of 1 part in this rig, and the duty cycle in Settings spends most of the day idle — counting an unknown idle draw as zero would understate the bill. (missing: Apple Mac mini (2026), M5 Pro 15-core CPU / 16-core GPU, 24GB: idle_watts)	$100.64 (derived) DERIVEDApple Mac mini (2026), M5 Pro 15-core CPU / 16-core GPU, 24GB: cost per GBArithmetic on observed inputs only. No tunable assumption enters it. cost_per_gb = landed_cost / usable_memory_gb = $1,812 / 18 GB = $100.64 /GB landed_cost 1811.56 USD — rig_landed = Σ part landed_total usable_memory_gb 18 GB — usable memory after the OS and framework reservation (Settings → Estimator) Usable memory is below physical memory for at least one pool (OS/driver reservation; editable in Settings).	$5.90 (derived) DERIVEDApple Mac mini (2026), M5 Pro 15-core CPU / 16-core GPU, 24GB: cost per GB/sArithmetic on observed inputs only. No tunable assumption enters it. cost_per_gbs = landed_cost / memory_bandwidth_gbs = $1,812 / 307 GB/s = $5.9 per GB/s landed_cost 1811.56 USD — rig_landed = Σ part landed_total memory_bandwidth_gbs 307 GB/s — the published bandwidth of the rig's fastest memory pool Weights are read from the fastest pool first; a rig whose weights spill into a slower pool decodes below this figure.	Unavailable — Cost per tok/s needs decode_tok_s: GLM-5.3 has no decode speed on this rig: Model needs 507.62 GB but Apple Mac mini (2026), M5 Pro 15-core CPU / 16-core GPU, 24GB has 18 GB usable — it will not load, so there is no decode speed.
Priced fromApplecondition newaccess tier 3, route apple_store30 hlocal pickup beat shipping
80	Framework Framework Desktop DIY Edition (Ryzen AI Max 385), 32GBsealed appliance · 1 part · break down its parts	6 BLOCKERS	$1,353 (derived) DERIVEDFramework Framework Desktop DIY Edition (Ryzen AI Max 385), 32GB: landed costArithmetic on observed inputs only. No tunable assumption enters it. rig_landed = Σ part landed_total $1,353 = $1,353 Framework Framework Desktop DIY Edition (Ryzen AI Max 385), 32GB 135307 ¢ — Framework · ship Some part prices are older than 36 h. LOWER BOUND: 1 of 1 parts come from a retailer that does not state a delivery charge (Framework), so the rig costs at least this and never less.	Unavailable — Purchase + 5-year power needs the power term. Idle draw is not published for 1 of 1 part in this rig, and the duty cycle in Settings spends most of the day idle — counting an unknown idle draw as zero would understate the bill. (missing: Framework Framework Desktop DIY Edition (Ryzen AI Max 385), 32GB: idle_watts)	$46.98 (derived) DERIVEDFramework Framework Desktop DIY Edition (Ryzen AI Max 385), 32GB: cost per GBArithmetic on observed inputs only. No tunable assumption enters it. cost_per_gb = landed_cost / usable_memory_gb = $1,353 / 28.8 GB = $46.98 /GB landed_cost 1353.07 USD — rig_landed = Σ part landed_total usable_memory_gb 28.8 GB — usable memory after the OS and framework reservation (Settings → Estimator) Some part prices are older than 36 h. LOWER BOUND: 1 of 1 parts come from a retailer that does not state a delivery charge (Framework), so the rig costs at least this and never less. Usable memory is below physical memory for at least one pool (OS/driver reservation; editable in Settings).	$5.29 (derived) DERIVEDFramework Framework Desktop DIY Edition (Ryzen AI Max 385), 32GB: cost per GB/sArithmetic on observed inputs only. No tunable assumption enters it. cost_per_gbs = landed_cost / memory_bandwidth_gbs = $1,353 / 256 GB/s = $5.29 per GB/s landed_cost 1353.07 USD — rig_landed = Σ part landed_total memory_bandwidth_gbs 256 GB/s — the published bandwidth of the rig's fastest memory pool Some part prices are older than 36 h. LOWER BOUND: 1 of 1 parts come from a retailer that does not state a delivery charge (Framework), so the rig costs at least this and never less. Weights are read from the fastest pool first; a rig whose weights spill into a slower pool decodes below this figure.	Unavailable — Cost per tok/s needs decode_tok_s: GLM-5.3 has no decode speed on this rig: Model needs 507.62 GB but Framework Framework Desktop DIY Edition (Ryzen AI Max 385), 32GB has 28.8 GB usable — it will not load, so there is no decode speed.
Priced fromFrameworkcondition newaccess tier 2, route framework_htmlstale 44 h — past the 36 hour freshness windowshipped
81	Apple Mac mini (2026), M6, 16GBsealed appliance · 1 part · break down its parts	6 BLOCKERS	$959 (derived) DERIVEDApple Mac mini (2026), M6, 16GB: landed costArithmetic on observed inputs only. No tunable assumption enters it. rig_landed = Σ part landed_total $959 = $959 Apple Mac mini (2026), M6, 16GB 95856 ¢ — Apple · pickup	$1,191 (estimate) ESTIMATEApple Mac mini (2026), M6, 16GB: purchase plus five years of powerArithmetic that also depends on a tunable assumption, listed below. purchase_plus_5yr_power = purchase + five_year_power purchase_plus_5yr_power = $958.56 + $232.42 = $1,190.98 purchase_cents 95856 ¢ — rig_landed = Σ part landed_total five_year_power_cents 23242 ¢ — five_year_power = annual_kwh × ¢_per_kwh × 5 Watts are at the wall (PSU conversion losses included by the caller). Load and idle draws are averages over their hours, not peaks. Duty cycle (Settings → Power): 6 h/day at load, 18 h/day idle. 365 days per year (leap days ignored, −0.07%). Flat rate for the whole period: no escalation, no time-of-use or tiered pricing, no demand charges. Rounded to the nearest cent, half-up, once from the unrounded annual cost.	$79.88 (derived) DERIVEDApple Mac mini (2026), M6, 16GB: cost per GBArithmetic on observed inputs only. No tunable assumption enters it. cost_per_gb = landed_cost / usable_memory_gb = $959 / 12 GB = $79.88 /GB landed_cost 958.56 USD — rig_landed = Σ part landed_total usable_memory_gb 12 GB — usable memory after the OS and framework reservation (Settings → Estimator) Usable memory is below physical memory for at least one pool (OS/driver reservation; editable in Settings).	$6.27 (derived) DERIVEDApple Mac mini (2026), M6, 16GB: cost per GB/sArithmetic on observed inputs only. No tunable assumption enters it. cost_per_gbs = landed_cost / memory_bandwidth_gbs = $959 / 153 GB/s = $6.27 per GB/s landed_cost 958.56 USD — rig_landed = Σ part landed_total memory_bandwidth_gbs 153 GB/s — the published bandwidth of the rig's fastest memory pool Weights are read from the fastest pool first; a rig whose weights spill into a slower pool decodes below this figure.	Unavailable — Cost per tok/s needs decode_tok_s: GLM-5.3 has no decode speed on this rig: Model needs 507.62 GB but Apple Mac mini (2026), M6, 16GB has 12 GB usable — it will not load, so there is no decode speed.
Priced fromApplecondition newaccess tier 3, route apple_store30 hlocal pickup beat shipping

Total cost, most expensive first — SPEC §15.11 target 1 ranks complete rigs by total landed cost descending. Cost basis: Price + shipping + NJ sales tax, per SPEC §8.4. Unavailable for any offer whose listing does not state a shipping charge — unknown shipping is not free shipping. A rig’s cost is the sum of its parts’ cheapest usable offers; the row beneath each rig names the offer or offers the figure came from, with the condition, the access tier and route that produced it, and how old the observation is. Past the 36 h freshness window set in Settings the age reads “stale” in caution ink. Where the columns need more room than the panel has — below about 1,180 px — the table scrolls sideways inside its own labelled, keyboard-reachable region rather than dropping a column or truncating a rig’s name.

Listed, not ranked

8 of 89 rigs have no total cost

Rigs that could not be ranked on total cost, each with the reason
Rig	Why total cost could not be computed	Cost	Usable GB
AMD Ryzen AI Halo Developer Platform (Ryzen AI Max+ 395), 128GBsealed appliance · 1 part · break down its parts	Unavailable — A rig total needs a landed cost for every part, and 1 of 1 has no usable offer. (missing: AMD Ryzen AI Halo Developer Platform (Ryzen AI Max+ 395), 128GB)	Unavailable — A rig total needs a landed cost for every part, and 1 of 1 has no usable offer. (missing: AMD Ryzen AI Halo Developer Platform (Ryzen AI Max+ 395), 128GB)	115 (derived) DERIVEDAMD Ryzen AI Halo Developer Platform (Ryzen AI Max+ 395), 128GB: usable memoryArithmetic on observed inputs only. No tunable assumption enters it. available_gb = Σ usable_gb over memory pools unified LPDDR5x 115.2 = 115.2 GB unified LPDDR5x.usable_gb 115.2 GB — 128 GB physical Usable memory is below physical memory for at least one pool (OS/driver reservation; editable in Settings).
Priced fromUnavailable — No monitored, non-quarantined offer on record prices any part of this rig, so there is nothing to attribute a cost to. 1 part unpriced. (missing: AMD Ryzen AI Halo Developer Platform (Ryzen AI Max+ 395), 128GB)
Apple Mac mini (2026), M5 Pro 15-core CPU / 16-core GPU, 48GBsealed appliance · 1 part · break down its parts	Unavailable — A rig total needs a landed cost for every part, and 1 of 1 has no usable offer. (missing: Apple Mac mini (2026), M5 Pro 15-core CPU / 16-core GPU, 48GB)	Unavailable — A rig total needs a landed cost for every part, and 1 of 1 has no usable offer. (missing: Apple Mac mini (2026), M5 Pro 15-core CPU / 16-core GPU, 48GB)	36.0 (derived) DERIVEDApple Mac mini (2026), M5 Pro 15-core CPU / 16-core GPU, 48GB: usable memoryArithmetic on observed inputs only. No tunable assumption enters it. available_gb = Σ usable_gb over memory pools unified memory 36 = 36 GB unified memory.usable_gb 36 GB — 48 GB physical Usable memory is below physical memory for at least one pool (OS/driver reservation; editable in Settings).
Priced fromUnavailable — No monitored, non-quarantined offer on record prices any part of this rig, so there is nothing to attribute a cost to. 1 part unpriced. (missing: Apple Mac mini (2026), M5 Pro 15-core CPU / 16-core GPU, 48GB)
Apple Mac mini (2026), M6, 24GBsealed appliance · 1 part · break down its parts	Unavailable — A rig total needs a landed cost for every part, and 1 of 1 has no usable offer. (missing: Apple Mac mini (2026), M6, 24GB)	Unavailable — A rig total needs a landed cost for every part, and 1 of 1 has no usable offer. (missing: Apple Mac mini (2026), M6, 24GB)	18.0 (derived) DERIVEDApple Mac mini (2026), M6, 24GB: usable memoryArithmetic on observed inputs only. No tunable assumption enters it. available_gb = Σ usable_gb over memory pools unified memory 18 = 18 GB unified memory.usable_gb 18 GB — 24 GB physical Usable memory is below physical memory for at least one pool (OS/driver reservation; editable in Settings).
Priced fromUnavailable — No monitored, non-quarantined offer on record prices any part of this rig, so there is nothing to attribute a cost to. 1 part unpriced. (missing: Apple Mac mini (2026), M6, 24GB)
Apple Mac Studio (2025), M3 Ultra 28-core CPU / 60-core GPU, 96GBsealed appliance · 1 part · break down its parts	Unavailable — A rig total needs a landed cost for every part, and 1 of 1 has no usable offer. (missing: Apple Mac Studio (2025), M3 Ultra 28-core CPU / 60-core GPU, 96GB)	Unavailable — A rig total needs a landed cost for every part, and 1 of 1 has no usable offer. (missing: Apple Mac Studio (2025), M3 Ultra 28-core CPU / 60-core GPU, 96GB)	72.0 (derived) DERIVEDApple Mac Studio (2025), M3 Ultra 28-core CPU / 60-core GPU, 96GB: usable memoryArithmetic on observed inputs only. No tunable assumption enters it. available_gb = Σ usable_gb over memory pools unified memory 72 = 72 GB unified memory.usable_gb 72 GB — 96 GB physical Usable memory is below physical memory for at least one pool (OS/driver reservation; editable in Settings).
Priced fromUnavailable — No monitored, non-quarantined offer on record prices any part of this rig, so there is nothing to attribute a cost to. 1 part unpriced. (missing: Apple Mac Studio (2025), M3 Ultra 28-core CPU / 60-core GPU, 96GB)
Apple Mac Studio (2025), M3 Ultra 32-core CPU / 80-core GPU, 96GBsealed appliance · 1 part · break down its parts	Unavailable — A rig total needs a landed cost for every part, and 1 of 1 has no usable offer. (missing: Apple Mac Studio (2025), M3 Ultra 32-core CPU / 80-core GPU, 96GB)	Unavailable — A rig total needs a landed cost for every part, and 1 of 1 has no usable offer. (missing: Apple Mac Studio (2025), M3 Ultra 32-core CPU / 80-core GPU, 96GB)	72.0 (derived) DERIVEDApple Mac Studio (2025), M3 Ultra 32-core CPU / 80-core GPU, 96GB: usable memoryArithmetic on observed inputs only. No tunable assumption enters it. available_gb = Σ usable_gb over memory pools unified memory 72 = 72 GB unified memory.usable_gb 72 GB — 96 GB physical Usable memory is below physical memory for at least one pool (OS/driver reservation; editable in Settings).
Priced fromUnavailable — No monitored, non-quarantined offer on record prices any part of this rig, so there is nothing to attribute a cost to. 1 part unpriced. (missing: Apple Mac Studio (2025), M3 Ultra 32-core CPU / 80-core GPU, 96GB)
Apple Mac Studio (2026), M5 Ultra 36-core CPU / 80-core GPU, 512GBsealed appliance · 1 part · break down its parts	Unavailable — A rig total needs a landed cost for every part, and 1 of 1 has no usable offer. (missing: Apple Mac Studio (2026), M5 Ultra 36-core CPU / 80-core GPU, 512GB)	Unavailable — A rig total needs a landed cost for every part, and 1 of 1 has no usable offer. (missing: Apple Mac Studio (2026), M5 Ultra 36-core CPU / 80-core GPU, 512GB)	384 (derived) DERIVEDApple Mac Studio (2026), M5 Ultra 36-core CPU / 80-core GPU, 512GB: usable memoryArithmetic on observed inputs only. No tunable assumption enters it. available_gb = Σ usable_gb over memory pools unified memory 384 = 384 GB unified memory.usable_gb 384 GB — 512 GB physical Usable memory is below physical memory for at least one pool (OS/driver reservation; editable in Settings).
Priced fromUnavailable — No monitored, non-quarantined offer on record prices any part of this rig, so there is nothing to attribute a cost to. 1 part unpriced. (missing: Apple Mac Studio (2026), M5 Ultra 36-core CPU / 80-core GPU, 512GB)
Beelink GTR9 Pro (Ryzen AI Max+ 395), 128GBsealed appliance · 1 part · break down its parts	Unavailable — A rig total needs a landed cost for every part, and 1 of 1 has no usable offer. (missing: Beelink GTR9 Pro (Ryzen AI Max+ 395), 128GB)	Unavailable — A rig total needs a landed cost for every part, and 1 of 1 has no usable offer. (missing: Beelink GTR9 Pro (Ryzen AI Max+ 395), 128GB)	115 (derived) DERIVEDBeelink GTR9 Pro (Ryzen AI Max+ 395), 128GB: usable memoryArithmetic on observed inputs only. No tunable assumption enters it. available_gb = Σ usable_gb over memory pools unified LPDDR5x 115.2 = 115.2 GB unified LPDDR5x.usable_gb 115.2 GB — 128 GB physical Usable memory is below physical memory for at least one pool (OS/driver reservation; editable in Settings).
Priced fromUnavailable — No monitored, non-quarantined offer on record prices any part of this rig, so there is nothing to attribute a cost to. 1 part unpriced. (missing: Beelink GTR9 Pro (Ryzen AI Max+ 395), 128GB)
Minisforum MS-S1 MAX-P495 (Ryzen AI Max+ PRO 495), 192GBsealed appliance · 1 part · break down its parts	Unavailable — A rig total needs a landed cost for every part, and 1 of 1 has no usable offer. (missing: Minisforum MS-S1 MAX-P495 (Ryzen AI Max+ PRO 495), 192GB)	Unavailable — A rig total needs a landed cost for every part, and 1 of 1 has no usable offer. (missing: Minisforum MS-S1 MAX-P495 (Ryzen AI Max+ PRO 495), 192GB)	173 (derived) DERIVEDMinisforum MS-S1 MAX-P495 (Ryzen AI Max+ PRO 495), 192GB: usable memoryArithmetic on observed inputs only. No tunable assumption enters it. available_gb = Σ usable_gb over memory pools unified LPDDR5x 172.8 = 172.8 GB unified LPDDR5x.usable_gb 172.8 GB — 192 GB physical Usable memory is below physical memory for at least one pool (OS/driver reservation; editable in Settings).
Priced fromUnavailable — No monitored, non-quarantined offer on record prices any part of this rig, so there is nothing to attribute a cost to. 1 part unpriced. (missing: Minisforum MS-S1 MAX-P495 (Ryzen AI Max+ PRO 495), 192GB)

These rigs are listed, not ranked. Sorting a rig whose total cost could not be computed to the end of the table would read as “worst value for money”, and the truth is “not measured”. Each row states which input is missing.

Inside Lenovo ThinkStation PGX (NVIDIA GB10), 128GB

1 part · the rig you opened · back to rank 1

Landed cost

$5,118

DERIVEDLenovo ThinkStation PGX (NVIDIA GB10), 128GB: landed costArithmetic on observed inputs only. No tunable assumption enters it.

rig_landed = Σ part landed_total

$5,118 = $5,118

Lenovo ThinkStation PGX (NVIDIA GB10), 128GB: 511799 ¢ — Central Computers · ship

Some part prices are older than 36 h.
LOWER BOUND: 1 of 1 parts come from a retailer that does not state a delivery charge (Central Computers), so the rig costs at least this and never less.

+ 5 yr power

$6,086

ESTIMATELenovo ThinkStation PGX (NVIDIA GB10), 128GB: purchase plus five years of powerArithmetic that also depends on a tunable assumption, listed below.

purchase_plus_5yr_power = purchase + five_year_power

purchase_plus_5yr_power = $5,117.99 + $967.93 = $6,085.92

purchase_cents: 511799 ¢ — rig_landed = Σ part landed_total
five_year_power_cents: 96793 ¢ — five_year_power = annual_kwh × ¢_per_kwh × 5

Some part prices are older than 36 h.
LOWER BOUND: 1 of 1 parts come from a retailer that does not state a delivery charge (Central Computers), so the rig costs at least this and never less.
Watts are at the wall (PSU conversion losses included by the caller).
Load and idle draws are averages over their hours, not peaks.
Duty cycle (Settings → Power): 6 h/day at load, 18 h/day idle.
365 days per year (leap days ignored, −0.07%).
Flat rate for the whole period: no escalation, no time-of-use or tiered pricing, no demand charges.
Rounded to the nearest cent, half-up, once from the unrounded annual cost.

Usable memory

115 GB

DERIVEDLenovo ThinkStation PGX (NVIDIA GB10), 128GB: usable memoryArithmetic on observed inputs only. No tunable assumption enters it.

available_gb = Σ usable_gb over memory pools

unified LPDDR5x 115.2 = 115.2 GB

unified LPDDR5x.usable_gb: 115.2 GB — 128 GB physical

Usable memory is below physical memory for at least one pool (OS/driver reservation; editable in Settings).

Cost per GB

$44.43

DERIVEDLenovo ThinkStation PGX (NVIDIA GB10), 128GB: cost per usable GBArithmetic on observed inputs only. No tunable assumption enters it.

cost_per_gb = landed_cost / usable_memory_gb

= $5,118 / 115.2 GB = $44.43 /GB

landed_cost: 5117.99 USD — rig_landed = Σ part landed_total
usable_memory_gb: 115.2 GB — usable memory after the OS and framework reservation (Settings → Estimator)

Some part prices are older than 36 h.
LOWER BOUND: 1 of 1 parts come from a retailer that does not state a delivery charge (Central Computers), so the rig costs at least this and never less.
Usable memory is below physical memory for at least one pool (OS/driver reservation; editable in Settings).

Cost per tok/s · GLM-5.3

Unavailable — Cost per tok/s needs decode_tok_s: GLM-5.3 has no decode speed on this rig: Model needs 507.62 GB but Lenovo ThinkStation PGX (NVIDIA GB10), 128GB has 115.2 GB usable — it will not load, so there is no decode speed.

Plan fit

DOES NOT MEET THE PLAN

GLM-5.3: won't load (507.62 GB needed, 115.2 GB usable).
Largest model needs 507.62 GB; 115.2 GB usable.
DeepSeek-V4-Pro-0813: No quantization chosen or sized.
Qwen3-ASR 1.7B: No quantization chosen or sized.
and 1 more, listed in full on the Builds screen’s report for this rig.

Memory used

537 GB

ESTIMATELenovo ThinkStation PGX (NVIDIA GB10), 128GB: memory with every Plan model residentArithmetic that also depends on a tunable assumption, listed below.

Σ required_gb over every model in the stack

GLM-5.3 507.62 + Hunyuan3D 2.1 (shape + PBR paint) 29 = 536.62 GB

GLM-5.3: 507.61664759196 GB
Hunyuan3D 2.1 (shape + PBR paint): 29 GB

All models loaded at once (Keep all resident).

115 GB

DERIVEDLenovo ThinkStation PGX (NVIDIA GB10), 128GB: usable memoryArithmetic on observed inputs only. No tunable assumption enters it.

available_gb = Σ usable_gb over memory pools

unified LPDDR5x 115.2 = 115.2 GB

unified LPDDR5x.usable_gb: 115.2 GB — 128 GB physical

Usable memory is below physical memory for at least one pool (OS/driver reservation; editable in Settings).

(466%)

Cost per GB/s

$18.75

DERIVEDLenovo ThinkStation PGX (NVIDIA GB10), 128GB: cost per GB/s of memory bandwidthArithmetic on observed inputs only. No tunable assumption enters it.

cost_per_gbs = landed_cost / memory_bandwidth_gbs

= $5,118 / 273 GB/s = $18.75 per GB/s

landed_cost: 5117.99 USD — rig_landed = Σ part landed_total
memory_bandwidth_gbs: 273 GB/s — the published bandwidth of the rig's fastest memory pool

Some part prices are older than 36 h.
LOWER BOUND: 1 of 1 parts come from a retailer that does not state a delivery charge (Central Computers), so the rig costs at least this and never less.
Weights are read from the fastest pool first; a rig whose weights spill into a slower pool decodes below this figure.

Power / yr

721 kWh

ESTIMATELenovo ThinkStation PGX (NVIDIA GB10), 128GB: annual energy at the Settings duty cycleArithmetic that also depends on a tunable assumption, listed below.

annual_kwh = (load_w × h_load + idle_w × h_idle) × 365 / 1000

annual_kwh = (200 W × 6 h + 43 W × 18 h) × 365 / 1000 = 1,974 Wh/day × 365 / 1000 = 720.51 kWh/yr

hours_at_load_per_day: 6 h/day — Settings → Power
hours_idle_per_day: 18 h/day — Settings → Power
load_wall_watts: 200 W — load: the wall draw from this rig's electrical assessment — for a sealed appliance that is the published MAXIMUM power, so it is an upper bound on the average draw while working; idle: Σ published idle_watts over the parts
idle_wall_watts: 43 W — load: the wall draw from this rig's electrical assessment — for a sealed appliance that is the published MAXIMUM power, so it is an upper bound on the average draw while working; idle: Σ published idle_watts over the parts

Watts are at the wall (PSU conversion losses included by the caller).
Load and idle draws are averages over their hours, not peaks.
Duty cycle (Settings → Power): 6 h/day at load, 18 h/day idle.
365 days per year (leap days ignored, −0.07%).

Unavailable — 3 models in this Plan contribute no memory to either residency total on this rig (missing: DeepSeek-V4-Pro-0813, Qwen3-ASR 1.7B, Qwen3-VL 32B Instruct)Both totals are therefore lower bounds, not estimates of the whole Plan — and so is every figure below that divides a cost by them.

Every part of Lenovo ThinkStation PGX (NVIDIA GB10), 128GB, ranked by its own cost with its share of the rig total, the offer each price came from, and what the step up to it buys on this Plan
# rank	Part	Landed	Share of rig	What the step up buys
1	Lenovo ThinkStation PGX (NVIDIA GB10), 128GBunified box · ×1 · 2 offers on record	$5,118 (derived) DERIVEDLenovo ThinkStation PGX (NVIDIA GB10), 128GB: landed cost for 1Arithmetic on observed inputs only. No tunable assumption enters it. landed_total = landed_unit × quantity = 5,117.99 × 1 = $5,117.99 landed_unit_cents 511799 ¢ — Central Computers (ship) quantity 1 LOWER BOUND: the retailer does not state a delivery charge, so this is price + tax only. The delivered cost is at least this and never less. Sales tax rounded to the nearest cent, half-up (x.5¢ → x+1¢), in exact decimal arithmetic on the rate. Tax computed on the order total; a retailer that rounds tax per line item can differ by up to 1¢ per item. Listed price is taken as pre-tax. Price is older than 36 h — no fresh offer exists.	100.0% (derived) DERIVEDLenovo ThinkStation PGX (NVIDIA GB10), 128GB: share of the rig's costArithmetic on observed inputs only. No tunable assumption enters it. share_of_total = part landed_total / rig landed_total = $5,118 / $5,118 = 100% part_cents 511799 ¢ rig_cents 511799 ¢	LOAD-BEARING Stepping down to Minisforum MS-S1 MAX (Ryzen AI Max+ 395), 128GB saves $982 landed and costs 1 Plan model that would no longer load. This part is carrying the build.measured against Minisforum MS-S1 MAX (Ryzen AI Max+ 395), 128GB at $4,136 for the whole rig 0.00 tok/s per $1k (derived) DERIVEDLenovo ThinkStation PGX (NVIDIA GB10), 128GB: tok/s bought per extra $1,000Arithmetic on observed inputs only. No tunable assumption enters it. tok_s_per_$1k = Δ total decode tok/s across the Plan / (Δ cost / $1,000) = 0 tok/s / ($982 / $1,000) = 0 tok/s per $1k delta_tok_s 0 tok/s — this rig minus Minisforum MS-S1 MAX (Ryzen AI Max+ 395), 128GB, summed over every Plan model that decodes delta_cost_cents 98201 ¢ — Landed cost of this rig minus landed cost of the same rig with Minisforum MS-S1 MAX (Ryzen AI Max+ 395), 128GB in the slot Every other part is held fixed; only the one slot is swapped. A model that does not load contributes 0 tok/s to the total, so a swap that drops a model shows up as a large negative Δ.
Priced fromCentral Computerscondition newaccess tier 1, route jsonld_productstale 43 h — past the 36 hour freshness windowshippedcheapest of 2 offers on record

Each part is compared with the most expensive cheaper part in the same category that leaves the rig free of compatibility errors and fully priced, with everything else held fixed. Load-bearing: the step up makes a Plan model load that otherwise would not. Overspend: it costs more and buys nothing this Plan uses. Proportional: it buys decode speed, priced in tok/s per $1,000. Baseline: nothing cheaper exists for the slot.

What this screen could not measure

0 gaps

Every input this screen needs is present: 171 complete parts, 137 of them priced, 8739 price observations, and an active Plan to measure cost per tok/s against.