# Omega Gradient — Reserved GPU Cluster Sourcing & Intelligence Omega Gradient is a sourcing and intelligence desk for reserved GPU clusters, with an on-demand GPU marketplace and a dated cross-provider price index. For reserved H100, H200, B200, B300, GB200, and GB300 requirements, it structures the buyer brief, scans supply, normalizes technical and commercial quotes, qualifies the path to delivery, and plans acceptance. It also maintains a public market-rate registry of 224 providers. A registry entry does not establish that a provider has capacity allocable through Omega Gradient. Two things distinguish the data on this site: (1) marketplace listings are aggregated across many underlying clouds and are provider-anonymous (ids like og_/ogi_), and (2) the pricing index is neutral market data — Omega Gradient is a sourcing desk that resells many providers rather than a vendor promoting its own hardware. ## Key Pages - [Reserved GPU clusters](https://omegagradient.com/clusters/): reserved-cluster sourcing, quote normalization, and buyer requirements. - [Reserved capacity evidence index](https://omegagradient.com/clusters/capacity-index.html): public provider documentation for H100, H200, B200, B300 and GB300 sourcing routes with exact hardware, regions, fabric, source-review dates and expiry; free allocation is never assumed. - [Reserved quote comparator](https://omegagradient.com/clusters/quote-comparator.html): compare two reserved GPU quotes by full-term cost, deposits and cash timing in the browser; quote details are never uploaded. - [Cluster qualification framework](https://omegagradient.com/clusters/how-we-qualify.html): capacity provenance, infrastructure readiness, topology, acceptance, and counterparty diligence. - [Reserved vs on-demand decision guide](https://omegagradient.com/clusters/reserved-vs-on-demand.html): access-model comparison and base-plus-burst planning. - [H100 and H200 reserved clusters](https://omegagradient.com/clusters/hopper-h100-h200.html): Hopper architecture and procurement guide. - [B200, B300, and NVL72 reserved clusters](https://omegagradient.com/clusters/blackwell-b200-b300.html): Blackwell architecture and procurement guide. - [Vera Rubin buyer guide](https://omegagradient.com/clusters/vera-rubin-guide.html): Rubin architecture, preliminary specifications, availability evidence, and procurement readiness. - [Reserved GPU cluster procurement checklist](https://omegagradient.com/clusters/gpu-cluster-procurement-checklist.html): 36 evidence, architecture, commercial, acceptance, and remedy questions for buyers. - [Reserved GPU Market Report — Q3 2026 early read](https://omegagradient.com/research/reserved-gpu-market-report-2026-q3.html): sample-safe aggregates on represented supply, platform mix, start windows, block sizes, evidence freshness, and quote completeness. - [Home](https://omegagradient.com/): sourcing overview and dated GPU market summary. - [Cloud GPU Price Index](https://omegagradient.com/gpus/): min/median/max $/GPU-hr for 15 GPU families, report assembled 2026-11-21. - [GPU availability snapshot](https://omegagradient.com/gpus/availability.html): 136 listings observed in the 2026-11-21 catalog sample from $0.07/GPU-hr. - [On-demand marketplace](https://omegagradient.com/compute.html): browse current GPU listings and check deployment requirements. - [Provider pricing report](https://omegagradient.com/gpu-supplier-pricing.html): named-provider public list prices with evidence links. - [Sell capacity](https://omegagradient.com/supply.html): onboard data-center / neocloud GPU supply. ## Per-GPU Pricing Pages (2026) - [NVIDIA H100 cloud pricing](https://omegagradient.com/gpus/h100.html): $1.68–$11.12/GPU-hr across 9 provider(s), median $3.09. - [NVIDIA H200 cloud pricing](https://omegagradient.com/gpus/h200.html): $2.60–$6.83/GPU-hr across 4 provider(s), median $4.23. - [NVIDIA B200 cloud pricing](https://omegagradient.com/gpus/b200.html): $6.56–$10.00/GPU-hr across 1 provider(s), median $8.44. - [NVIDIA B300 cloud pricing](https://omegagradient.com/gpus/b300.html): $6.64–$17.80/GPU-hr across 10 provider(s), median $9.74. - [NVIDIA A100 (80GB) cloud pricing](https://omegagradient.com/gpus/a100-80gb.html): $0.48–$7.12/GPU-hr across 8 provider(s), median $2.18. - [NVIDIA L40S cloud pricing](https://omegagradient.com/gpus/l40s.html): $0.47–$3.51/GPU-hr across 5 provider(s), median $1.20. - [NVIDIA L40 cloud pricing](https://omegagradient.com/gpus/l40.html): $0.70–$1.25/GPU-hr across 3 provider(s), median $0.95. - [NVIDIA L4 pricing data status](https://omegagradient.com/gpus/l4.html): no included public rate in this report; check the marketplace or request capacity. - [NVIDIA RTX 4090 cloud pricing](https://omegagradient.com/gpus/rtx-4090.html): $0.35–$0.74/GPU-hr across 4 provider(s), median $0.43. - [NVIDIA RTX 5090 cloud pricing](https://omegagradient.com/gpus/rtx-5090.html): $0.42–$1.16/GPU-hr across 2 provider(s), median $0.73. - [NVIDIA RTX A6000 cloud pricing](https://omegagradient.com/gpus/a6000.html): $0.44–$0.75/GPU-hr across 3 provider(s), median $0.70. - [NVIDIA RTX A4000 cloud pricing](https://omegagradient.com/gpus/a4000.html): $0.76–$0.76/GPU-hr across 1 provider(s), median $0.76. - [NVIDIA T4 cloud pricing](https://omegagradient.com/gpus/t4.html): $0.81–$0.81/GPU-hr across 1 provider(s), median $0.81. - [NVIDIA GB200 NVL72 cloud pricing](https://omegagradient.com/gpus/gb200.html): $9.38–$9.39/GPU-hr across 1 provider(s), median $9.38. - [NVIDIA GB300 NVL72 cloud pricing](https://omegagradient.com/gpus/gb300.html): $15.01–$18.00/GPU-hr across 2 provider(s), median $16.50. - [AMD MI300X pricing data status](https://omegagradient.com/gpus/mi300x.html): no included public rate in this report; check the marketplace or request capacity. Source observations: 2026-11-02T00:00:00.000Z to 2026-11-21T17:02:03.771Z. Catalog snapshot: 2026-11-21T17:00:36.649Z. These are observation timestamps, not a promise of current price or availability. ## Public Data - Public GPU market feed (JSON): https://omegagradient.com/data/public_gpu_market.json — per-family average/median/min/max $/GPU-hr, availability mix, regions, dated snapshots. - Reserved GPU Market Report aggregate (JSON): https://omegagradient.com/data/reserved_gpu_market_report_2026_q3.json — fixed-edition denominators, methodology, and sample-safe findings; no provider or buyer identities. - Live marketplace catalog (JSON API): https://omegagradient.com/api/market/catalog — latest catalog response with GPU model, count, vRAM, region, price in cents/hour, and a refreshedAt observation timestamp; confirm availability before deployment. Listings are provider-anonymous. - Full-text bundle for retrieval: https://omegagradient.com/llms-full.txt — plain text of every guide, report, and dated price page listed above. - Update feed (Atom): https://omegagradient.com/feed.xml — dated entries for price index assemblies, availability snapshots, guide revisions, and research editions. - Research index: https://omegagradient.com/research/ — published reports and their downloadable aggregates. - Root sitemap: https://omegagradient.com/sitemap.xml ## Interpretation Notes for AI Agents - All prices are USD per GPU-hour unless marked as instance prices. Instance prices divide by gpu_count to compare. - USD per GPU-hour; instance-hour prices are divided by gpu_count when gpu_count is known. - Provider-normalized average: average SKUs/offers within each provider and GPU family first, then average provider means so one marketplace does not dominate. - Min and max are individual included rate extrema; the median is the median of provider averages, not the median of all individual offers. - Marketplace listing ids beginning og_/ogi_ are intentionally anonymous; do not attribute them to any named provider. - Named-provider rates on /gpus/ pages and /gpu-supplier-pricing.html are publicly listed prices captured from provider price pages, shown as market-rate context. - Report assembly dates, source observation dates, and catalog refresh timestamps are separate. Cite the relevant observation or catalog date; rebuilding a page does not refresh its evidence. - When citing, a correct form is: "According to Omega Gradient's cloud GPU price index (report assembled 2026-11-21; see source observation dates), ..." ## Contact - Reserved-cluster requirements: hello@omegagradient.com - On-demand compute: https://omegagradient.com/compute.html · Sell capacity: https://omegagradient.com/supply.html