| { |
| "captured_at": "2026-07-27T05:22:34+00:00", |
| "comparison_policy": { |
| "benchmark_transfer": "Prices outside RunPod are projections using measured RunPod TPS, not measurements on the quoted provider.", |
| "excluded_costs": [ |
| "persistent storage unless inseparable from the instance", |
| "network egress", |
| "tax", |
| "setup and idle time", |
| "commitment discounts" |
| ], |
| "included_costs": "advertised accelerator or bundled-instance rate", |
| "vast_quote_rule": "Median of the three cheapest verified, rentable, single-GPU, on-demand offers with reliability >=0.98, CUDA >=12.8, and VRAM matching the measured SKU.", |
| "warning": "Provider products differ in CPU, RAM, storage, networking, availability, tenancy, and minimum rental commitment." |
| }, |
| "currency": "USD", |
| "quotes": [ |
| { |
| "gpu": "NVIDIA A100 80GB PCIe", |
| "instance_gpu_count": 8, |
| "match_quality": "near", |
| "notes": "A100 80 GB; instance fabric/clock may differ from measured PCIe host.", |
| "price_type": "capacity_block", |
| "product": "p4de.24xlarge", |
| "provider": "AWS", |
| "region": "us-east-1", |
| "source_url": "https://aws.amazon.com/ec2/capacityblocks/pricing/", |
| "usd_per_gpu_hour": 2.214 |
| }, |
| { |
| "gpu": "NVIDIA A100 80GB PCIe", |
| "instance_gpu_count": 1, |
| "match_quality": "near", |
| "notes": "Full A100 80 GB VM; interconnect/clock may differ from measured PCIe host.", |
| "price_type": "on_demand", |
| "product": "a2-ultragpu-1g", |
| "provider": "Google Cloud", |
| "region": "us-central1", |
| "source_url": "https://cloud.google.com/products/compute/pricing/accelerator-optimized", |
| "usd_per_gpu_hour": 5.06879789 |
| }, |
| { |
| "gpu": "NVIDIA A100 80GB PCIe", |
| "instance_gpu_count": 8, |
| "match_quality": "near", |
| "notes": "Measured device is PCIe; Lambda quote is an 8-GPU SXM instance.", |
| "price_type": "on_demand", |
| "product": "8x A100 SXM 80 GB", |
| "provider": "Lambda", |
| "region": null, |
| "source_url": "https://lambda.ai/instances", |
| "usd_per_gpu_hour": 2.79 |
| }, |
| { |
| "gpu": "NVIDIA A100 80GB PCIe", |
| "instance_gpu_count": 1, |
| "match_quality": "exact", |
| "notes": "", |
| "price_type": "marketplace_on_demand", |
| "product": "A100 PCIe 80GB", |
| "provider": "RunPod Community", |
| "region": null, |
| "source_url": "https://www.runpod.io/pricing", |
| "usd_per_gpu_hour": 1.19 |
| }, |
| { |
| "gpu": "NVIDIA A100 80GB PCIe", |
| "instance_gpu_count": 1, |
| "match_quality": "exact", |
| "notes": "", |
| "price_type": "on_demand", |
| "product": "A100 PCIe 80GB", |
| "provider": "RunPod Secure", |
| "region": null, |
| "source_url": "https://www.runpod.io/pricing", |
| "usd_per_gpu_hour": 1.39 |
| }, |
| { |
| "gpu": "NVIDIA A100 80GB PCIe", |
| "instance_gpu_count": 1, |
| "match_quality": "exact", |
| "notes": "Median of 3 cheapest eligible offers; range $0.867--$1.000/GPU-hour.", |
| "price_type": "marketplace_on_demand", |
| "product": "A100 PCIE", |
| "provider": "Vast.ai", |
| "region": null, |
| "source_url": "https://docs.vast.ai/api-reference/search/search-offers", |
| "usd_per_gpu_hour": 0.988 |
| }, |
| { |
| "gpu": "NVIDIA B200", |
| "instance_gpu_count": 8, |
| "match_quality": "exact", |
| "notes": "", |
| "price_type": "capacity_block", |
| "product": "p6-b200.48xlarge", |
| "provider": "AWS", |
| "region": "us-east-1", |
| "source_url": "https://aws.amazon.com/ec2/capacityblocks/pricing/", |
| "usd_per_gpu_hour": 12.355 |
| }, |
| { |
| "gpu": "NVIDIA B200", |
| "instance_gpu_count": 8, |
| "match_quality": "exact", |
| "notes": "No ordinary on-demand price was listed at capture time.", |
| "price_type": "flex_start", |
| "product": "a4-highgpu-8g", |
| "provider": "Google Cloud", |
| "region": "us-central1", |
| "source_url": "https://cloud.google.com/products/compute/pricing/accelerator-optimized", |
| "usd_per_gpu_hour": 8.055 |
| }, |
| { |
| "gpu": "NVIDIA B200", |
| "instance_gpu_count": 1, |
| "match_quality": "exact", |
| "notes": "", |
| "price_type": "on_demand", |
| "product": "1x B200 SXM6 180 GB", |
| "provider": "Lambda", |
| "region": null, |
| "source_url": "https://lambda.ai/instances", |
| "usd_per_gpu_hour": 6.99 |
| }, |
| { |
| "gpu": "NVIDIA B200", |
| "instance_gpu_count": 1, |
| "match_quality": "exact", |
| "notes": "", |
| "price_type": "marketplace_on_demand", |
| "product": "B200 180GB", |
| "provider": "RunPod Community", |
| "region": null, |
| "source_url": "https://www.runpod.io/pricing", |
| "usd_per_gpu_hour": 5.98 |
| }, |
| { |
| "gpu": "NVIDIA B200", |
| "instance_gpu_count": 1, |
| "match_quality": "exact", |
| "notes": "", |
| "price_type": "on_demand", |
| "product": "B200 180GB", |
| "provider": "RunPod Secure", |
| "region": null, |
| "source_url": "https://www.runpod.io/pricing", |
| "usd_per_gpu_hour": 5.89 |
| }, |
| { |
| "gpu": "NVIDIA B200", |
| "instance_gpu_count": 1, |
| "match_quality": "exact", |
| "notes": "Median of 3 cheapest eligible offers; range $6.237--$6.487/GPU-hour.", |
| "price_type": "marketplace_on_demand", |
| "product": "B200", |
| "provider": "Vast.ai", |
| "region": null, |
| "source_url": "https://docs.vast.ai/api-reference/search/search-offers", |
| "usd_per_gpu_hour": 6.25 |
| }, |
| { |
| "gpu": "NVIDIA B300 SXM6 AC", |
| "instance_gpu_count": 8, |
| "match_quality": "exact", |
| "notes": "", |
| "price_type": "capacity_block", |
| "product": "p6-b300.48xlarge", |
| "provider": "AWS", |
| "region": "us-east-1", |
| "source_url": "https://aws.amazon.com/ec2/capacityblocks/pricing/", |
| "usd_per_gpu_hour": 14.04 |
| }, |
| { |
| "gpu": "NVIDIA B300 SXM6 AC", |
| "instance_gpu_count": 1, |
| "match_quality": "exact", |
| "notes": "", |
| "price_type": "marketplace_on_demand", |
| "product": "B300 288GB", |
| "provider": "RunPod Community", |
| "region": null, |
| "source_url": "https://www.runpod.io/pricing", |
| "usd_per_gpu_hour": 6.94 |
| }, |
| { |
| "gpu": "NVIDIA B300 SXM6 AC", |
| "instance_gpu_count": 1, |
| "match_quality": "exact", |
| "notes": "", |
| "price_type": "on_demand", |
| "product": "B300 288GB", |
| "provider": "RunPod Secure", |
| "region": null, |
| "source_url": "https://www.runpod.io/pricing", |
| "usd_per_gpu_hour": 7.39 |
| }, |
| { |
| "gpu": "NVIDIA GeForce RTX 4090", |
| "instance_gpu_count": 1, |
| "match_quality": "exact", |
| "notes": "", |
| "price_type": "marketplace_on_demand", |
| "product": "RTX 4090", |
| "provider": "RunPod Community", |
| "region": null, |
| "source_url": "https://www.runpod.io/pricing", |
| "usd_per_gpu_hour": 0.34 |
| }, |
| { |
| "gpu": "NVIDIA GeForce RTX 4090", |
| "instance_gpu_count": 1, |
| "match_quality": "exact", |
| "notes": "", |
| "price_type": "on_demand", |
| "product": "RTX 4090", |
| "provider": "RunPod Secure", |
| "region": null, |
| "source_url": "https://www.runpod.io/pricing", |
| "usd_per_gpu_hour": 0.69 |
| }, |
| { |
| "gpu": "NVIDIA GeForce RTX 4090", |
| "instance_gpu_count": 1, |
| "match_quality": "exact", |
| "notes": "Median of 3 cheapest eligible offers; range $0.253--$0.253/GPU-hour.", |
| "price_type": "marketplace_on_demand", |
| "product": "RTX 4090", |
| "provider": "Vast.ai", |
| "region": null, |
| "source_url": "https://docs.vast.ai/api-reference/search/search-offers", |
| "usd_per_gpu_hour": 0.2533333333333333 |
| }, |
| { |
| "gpu": "NVIDIA GeForce RTX 5090", |
| "instance_gpu_count": 1, |
| "match_quality": "exact", |
| "notes": "", |
| "price_type": "marketplace_on_demand", |
| "product": "RTX 5090", |
| "provider": "RunPod Community", |
| "region": null, |
| "source_url": "https://www.runpod.io/pricing", |
| "usd_per_gpu_hour": 0.69 |
| }, |
| { |
| "gpu": "NVIDIA GeForce RTX 5090", |
| "instance_gpu_count": 1, |
| "match_quality": "exact", |
| "notes": "", |
| "price_type": "on_demand", |
| "product": "RTX 5090", |
| "provider": "RunPod Secure", |
| "region": null, |
| "source_url": "https://www.runpod.io/pricing", |
| "usd_per_gpu_hour": 0.99 |
| }, |
| { |
| "gpu": "NVIDIA GeForce RTX 5090", |
| "instance_gpu_count": 1, |
| "match_quality": "exact", |
| "notes": "Median of 3 cheapest eligible offers; range $0.280--$0.307/GPU-hour.", |
| "price_type": "marketplace_on_demand", |
| "product": "RTX 5090", |
| "provider": "Vast.ai", |
| "region": null, |
| "source_url": "https://docs.vast.ai/api-reference/search/search-offers", |
| "usd_per_gpu_hour": 0.29333333333333333 |
| }, |
| { |
| "gpu": "NVIDIA H100 80GB HBM3", |
| "instance_gpu_count": 1, |
| "match_quality": "exact", |
| "notes": "Single-H100 Capacity Block effective hourly rate.", |
| "price_type": "capacity_block", |
| "product": "p5.4xlarge", |
| "provider": "AWS", |
| "region": "us-east-1", |
| "source_url": "https://aws.amazon.com/ec2/capacityblocks/pricing/", |
| "usd_per_gpu_hour": 5.191 |
| }, |
| { |
| "gpu": "NVIDIA H100 80GB HBM3", |
| "instance_gpu_count": 8, |
| "match_quality": "exact", |
| "notes": "", |
| "price_type": "on_demand", |
| "product": "a3-highgpu-8g", |
| "provider": "Google Cloud", |
| "region": "us-central1", |
| "source_url": "https://cloud.google.com/products/compute/pricing/accelerator-optimized", |
| "usd_per_gpu_hour": 11.061250014875 |
| }, |
| { |
| "gpu": "NVIDIA H100 80GB HBM3", |
| "instance_gpu_count": 1, |
| "match_quality": "exact", |
| "notes": "", |
| "price_type": "on_demand", |
| "product": "1x H100 SXM 80 GB", |
| "provider": "Lambda", |
| "region": null, |
| "source_url": "https://lambda.ai/instances", |
| "usd_per_gpu_hour": 4.29 |
| }, |
| { |
| "gpu": "NVIDIA H100 80GB HBM3", |
| "instance_gpu_count": 1, |
| "match_quality": "exact", |
| "notes": "", |
| "price_type": "marketplace_on_demand", |
| "product": "H100 SXM 80GB", |
| "provider": "RunPod Community", |
| "region": null, |
| "source_url": "https://www.runpod.io/pricing", |
| "usd_per_gpu_hour": 2.69 |
| }, |
| { |
| "gpu": "NVIDIA H100 80GB HBM3", |
| "instance_gpu_count": 1, |
| "match_quality": "exact", |
| "notes": "", |
| "price_type": "on_demand", |
| "product": "H100 SXM 80GB", |
| "provider": "RunPod Secure", |
| "region": null, |
| "source_url": "https://www.runpod.io/pricing", |
| "usd_per_gpu_hour": 2.99 |
| }, |
| { |
| "gpu": "NVIDIA H100 80GB HBM3", |
| "instance_gpu_count": 1, |
| "match_quality": "exact", |
| "notes": "Median of 3 cheapest eligible offers; range $2.267--$3.000/GPU-hour.", |
| "price_type": "marketplace_on_demand", |
| "product": "H100 SXM", |
| "provider": "Vast.ai", |
| "region": null, |
| "source_url": "https://docs.vast.ai/api-reference/search/search-offers", |
| "usd_per_gpu_hour": 2.5453333333333332 |
| }, |
| { |
| "gpu": "NVIDIA H200", |
| "instance_gpu_count": 8, |
| "match_quality": "exact", |
| "notes": "", |
| "price_type": "capacity_block", |
| "product": "p5e.48xlarge", |
| "provider": "AWS", |
| "region": "us-east-2", |
| "source_url": "https://aws.amazon.com/ec2/capacityblocks/pricing/", |
| "usd_per_gpu_hour": 5.97 |
| }, |
| { |
| "gpu": "NVIDIA H200", |
| "instance_gpu_count": 8, |
| "match_quality": "exact", |
| "notes": "", |
| "price_type": "on_demand", |
| "product": "a3-ultragpu-8g", |
| "provider": "Google Cloud", |
| "region": "us-central1", |
| "source_url": "https://cloud.google.com/products/compute/pricing/accelerator-optimized", |
| "usd_per_gpu_hour": 10.600863561625 |
| }, |
| { |
| "gpu": "NVIDIA H200", |
| "instance_gpu_count": 1, |
| "match_quality": "exact", |
| "notes": "", |
| "price_type": "marketplace_on_demand", |
| "product": "H200 SXM 141GB", |
| "provider": "RunPod Community", |
| "region": null, |
| "source_url": "https://www.runpod.io/pricing", |
| "usd_per_gpu_hour": 3.59 |
| }, |
| { |
| "gpu": "NVIDIA H200", |
| "instance_gpu_count": 1, |
| "match_quality": "exact", |
| "notes": "", |
| "price_type": "on_demand", |
| "product": "H200 SXM 141GB", |
| "provider": "RunPod Secure", |
| "region": null, |
| "source_url": "https://www.runpod.io/pricing", |
| "usd_per_gpu_hour": 4.39 |
| }, |
| { |
| "gpu": "NVIDIA H200", |
| "instance_gpu_count": 1, |
| "match_quality": "exact", |
| "notes": "Median of 3 cheapest eligible offers; range $3.934--$4.287/GPU-hour.", |
| "price_type": "marketplace_on_demand", |
| "product": "H200", |
| "provider": "Vast.ai", |
| "region": null, |
| "source_url": "https://docs.vast.ai/api-reference/search/search-offers", |
| "usd_per_gpu_hour": 4.078947368421053 |
| }, |
| { |
| "gpu": "NVIDIA RTX 6000 Ada Generation", |
| "instance_gpu_count": 1, |
| "match_quality": "exact", |
| "notes": "", |
| "price_type": "marketplace_on_demand", |
| "product": "RTX 6000 Ada", |
| "provider": "RunPod Community", |
| "region": null, |
| "source_url": "https://www.runpod.io/pricing", |
| "usd_per_gpu_hour": 0.74 |
| }, |
| { |
| "gpu": "NVIDIA RTX 6000 Ada Generation", |
| "instance_gpu_count": 1, |
| "match_quality": "exact", |
| "notes": "", |
| "price_type": "on_demand", |
| "product": "RTX 6000 Ada", |
| "provider": "RunPod Secure", |
| "region": null, |
| "source_url": "https://www.runpod.io/pricing", |
| "usd_per_gpu_hour": 0.84 |
| }, |
| { |
| "gpu": "NVIDIA RTX 6000 Ada Generation", |
| "instance_gpu_count": 1, |
| "match_quality": "exact", |
| "notes": "Median of 3 cheapest eligible offers; range $0.533--$0.560/GPU-hour.", |
| "price_type": "marketplace_on_demand", |
| "product": "RTX 6000Ada", |
| "provider": "Vast.ai", |
| "region": null, |
| "source_url": "https://docs.vast.ai/api-reference/search/search-offers", |
| "usd_per_gpu_hour": 0.5479999999999999 |
| }, |
| { |
| "gpu": "NVIDIA RTX A6000", |
| "instance_gpu_count": 1, |
| "match_quality": "exact", |
| "notes": "", |
| "price_type": "on_demand", |
| "product": "1x RTX A6000", |
| "provider": "Lambda", |
| "region": null, |
| "source_url": "https://lambda.ai/instances", |
| "usd_per_gpu_hour": 1.09 |
| }, |
| { |
| "gpu": "NVIDIA RTX A6000", |
| "instance_gpu_count": 1, |
| "match_quality": "exact", |
| "notes": "", |
| "price_type": "marketplace_on_demand", |
| "product": "RTX A6000", |
| "provider": "RunPod Community", |
| "region": null, |
| "source_url": "https://www.runpod.io/pricing", |
| "usd_per_gpu_hour": 0.33 |
| }, |
| { |
| "gpu": "NVIDIA RTX A6000", |
| "instance_gpu_count": 1, |
| "match_quality": "exact", |
| "notes": "", |
| "price_type": "on_demand", |
| "product": "RTX A6000", |
| "provider": "RunPod Secure", |
| "region": null, |
| "source_url": "https://www.runpod.io/pricing", |
| "usd_per_gpu_hour": 0.53 |
| }, |
| { |
| "gpu": "NVIDIA RTX A6000", |
| "instance_gpu_count": 1, |
| "match_quality": "exact", |
| "notes": "Median of 1 cheapest eligible offers; range $0.387--$0.387/GPU-hour.", |
| "price_type": "marketplace_on_demand", |
| "product": "RTX A6000", |
| "provider": "Vast.ai", |
| "region": null, |
| "source_url": "https://docs.vast.ai/api-reference/search/search-offers", |
| "usd_per_gpu_hour": 0.38666666666666666 |
| }, |
| { |
| "gpu": "NVIDIA RTX PRO 6000 Blackwell Server Edition", |
| "instance_gpu_count": 1, |
| "match_quality": "exact", |
| "notes": "", |
| "price_type": "on_demand", |
| "product": "g4-standard-48", |
| "provider": "Google Cloud", |
| "region": "us-central1", |
| "source_url": "https://cloud.google.com/products/compute/pricing/accelerator-optimized", |
| "usd_per_gpu_hour": 4.49993 |
| }, |
| { |
| "gpu": "NVIDIA RTX PRO 6000 Blackwell Server Edition", |
| "instance_gpu_count": 1, |
| "match_quality": "exact", |
| "notes": "", |
| "price_type": "marketplace_on_demand", |
| "product": "RTX PRO 6000", |
| "provider": "RunPod Community", |
| "region": null, |
| "source_url": "https://www.runpod.io/pricing", |
| "usd_per_gpu_hour": 1.69 |
| }, |
| { |
| "gpu": "NVIDIA RTX PRO 6000 Blackwell Server Edition", |
| "instance_gpu_count": 1, |
| "match_quality": "exact", |
| "notes": "", |
| "price_type": "on_demand", |
| "product": "RTX PRO 6000", |
| "provider": "RunPod Secure", |
| "region": null, |
| "source_url": "https://www.runpod.io/pricing", |
| "usd_per_gpu_hour": 1.99 |
| }, |
| { |
| "gpu": "NVIDIA RTX PRO 6000 Blackwell Server Edition", |
| "instance_gpu_count": 1, |
| "match_quality": "exact", |
| "notes": "Median of 3 cheapest eligible offers; range $1.267--$1.333/GPU-hour.", |
| "price_type": "marketplace_on_demand", |
| "product": "RTX PRO 6000 S", |
| "provider": "Vast.ai", |
| "region": null, |
| "source_url": "https://docs.vast.ai/api-reference/search/search-offers", |
| "usd_per_gpu_hour": 1.2666666666666666 |
| } |
| ], |
| "schema_version": 1, |
| "sources": [ |
| "https://aws.amazon.com/ec2/capacityblocks/pricing/", |
| "https://cloud.google.com/products/compute/pricing/accelerator-optimized", |
| "https://docs.vast.ai/api-reference/search/search-offers", |
| "https://lambda.ai/instances", |
| "https://www.runpod.io/pricing" |
| ], |
| "unit": "per GPU-hour", |
| "vast_query_audit": { |
| "NVIDIA A100 80GB PCIe": { |
| "cheapest_three_sanitized": [ |
| { |
| "country_code": "US", |
| "cuda_max_good": 13.0, |
| "reliability": 0.9981, |
| "total_usd_per_hour_at_listed_storage": 0.8733333333333334, |
| "usd_per_gpu_hour": 0.8666666666666667, |
| "vram_mb": 81920.0 |
| }, |
| { |
| "country_code": "JP", |
| "cuda_max_good": 13.3, |
| "reliability": 0.992683, |
| "total_usd_per_hour_at_listed_storage": 0.9914311111111112, |
| "usd_per_gpu_hour": 0.988, |
| "vram_mb": 81920.0 |
| }, |
| { |
| "country_code": "GB", |
| "cuda_max_good": 13.1, |
| "reliability": 0.9896447, |
| "total_usd_per_hour_at_listed_storage": 1.0022222222222221, |
| "usd_per_gpu_hour": 1.0, |
| "vram_mb": 81920.0 |
| } |
| ], |
| "eligible_offer_count": 6, |
| "query_gpu_name": "A100 PCIE", |
| "quote_rule": "median of three cheapest eligible offers" |
| }, |
| "NVIDIA B200": { |
| "cheapest_three_sanitized": [ |
| { |
| "country_code": "US", |
| "cuda_max_good": 13.2, |
| "reliability": 0.9976631, |
| "total_usd_per_hour_at_listed_storage": 6.247916666666667, |
| "usd_per_gpu_hour": 6.2375, |
| "vram_mb": 183359.0 |
| }, |
| { |
| "country_code": "US", |
| "cuda_max_good": 13.0, |
| "reliability": 0.9981403, |
| "total_usd_per_hour_at_listed_storage": 6.252083333333333, |
| "usd_per_gpu_hour": 6.25, |
| "vram_mb": 183359.0 |
| }, |
| { |
| "country_code": "US", |
| "cuda_max_good": 13.2, |
| "reliability": 0.9957008, |
| "total_usd_per_hour_at_listed_storage": 6.497916666666667, |
| "usd_per_gpu_hour": 6.4875, |
| "vram_mb": 183359.0 |
| } |
| ], |
| "eligible_offer_count": 5, |
| "query_gpu_name": "B200", |
| "quote_rule": "median of three cheapest eligible offers" |
| }, |
| "NVIDIA B300 SXM6 AC": { |
| "cheapest_three_sanitized": [], |
| "eligible_offer_count": 0, |
| "query_gpu_name": "B300", |
| "quote_rule": "median of three cheapest eligible offers" |
| }, |
| "NVIDIA GeForce RTX 4090": { |
| "cheapest_three_sanitized": [ |
| { |
| "country_code": "US", |
| "cuda_max_good": 13.0, |
| "reliability": 0.9955969, |
| "total_usd_per_hour_at_listed_storage": 0.26296296296296295, |
| "usd_per_gpu_hour": 0.2533333333333333, |
| "vram_mb": 24564.0 |
| }, |
| { |
| "country_code": "US", |
| "cuda_max_good": 13.0, |
| "reliability": 0.9961348, |
| "total_usd_per_hour_at_listed_storage": 0.26296296296296295, |
| "usd_per_gpu_hour": 0.2533333333333333, |
| "vram_mb": 24564.0 |
| }, |
| { |
| "country_code": "US", |
| "cuda_max_good": 13.0, |
| "reliability": 0.9933616, |
| "total_usd_per_hour_at_listed_storage": 0.26296296296296295, |
| "usd_per_gpu_hour": 0.2533333333333333, |
| "vram_mb": 24564.0 |
| } |
| ], |
| "eligible_offer_count": 38, |
| "query_gpu_name": "RTX 4090", |
| "quote_rule": "median of three cheapest eligible offers" |
| }, |
| "NVIDIA GeForce RTX 5090": { |
| "cheapest_three_sanitized": [ |
| { |
| "country_code": "US", |
| "cuda_max_good": 13.0, |
| "reliability": 0.9933666, |
| "total_usd_per_hour_at_listed_storage": 0.2896296296296297, |
| "usd_per_gpu_hour": 0.28, |
| "vram_mb": 32607.0 |
| }, |
| { |
| "country_code": "US", |
| "cuda_max_good": 13.0, |
| "reliability": 0.9945708, |
| "total_usd_per_hour_at_listed_storage": 0.302962962962963, |
| "usd_per_gpu_hour": 0.29333333333333333, |
| "vram_mb": 32607.0 |
| }, |
| { |
| "country_code": "US", |
| "cuda_max_good": 13.0, |
| "reliability": 0.9949772, |
| "total_usd_per_hour_at_listed_storage": 0.30814814814814817, |
| "usd_per_gpu_hour": 0.3066666666666667, |
| "vram_mb": 32607.0 |
| } |
| ], |
| "eligible_offer_count": 55, |
| "query_gpu_name": "RTX 5090", |
| "quote_rule": "median of three cheapest eligible offers" |
| }, |
| "NVIDIA H100 80GB HBM3": { |
| "cheapest_three_sanitized": [ |
| { |
| "country_code": "MY", |
| "cuda_max_good": 13.2, |
| "reliability": 0.9905361, |
| "total_usd_per_hour_at_listed_storage": 2.2696296296296294, |
| "usd_per_gpu_hour": 2.2666666666666666, |
| "vram_mb": 81559.0 |
| }, |
| { |
| "country_code": "CZ", |
| "cuda_max_good": 13.2, |
| "reliability": 0.9958913, |
| "total_usd_per_hour_at_listed_storage": 2.551851851851852, |
| "usd_per_gpu_hour": 2.5453333333333332, |
| "vram_mb": 81559.0 |
| }, |
| { |
| "country_code": "TW", |
| "cuda_max_good": 13.0, |
| "reliability": 0.9834368, |
| "total_usd_per_hour_at_listed_storage": 3.005333333333333, |
| "usd_per_gpu_hour": 3.0, |
| "vram_mb": 81559.0 |
| } |
| ], |
| "eligible_offer_count": 8, |
| "query_gpu_name": "H100 SXM", |
| "quote_rule": "median of three cheapest eligible offers" |
| }, |
| "NVIDIA H200": { |
| "cheapest_three_sanitized": [ |
| { |
| "country_code": "SA", |
| "cuda_max_good": 13.0, |
| "reliability": 0.9974872, |
| "total_usd_per_hour_at_listed_storage": 3.9356725146198834, |
| "usd_per_gpu_hour": 3.93421052631579, |
| "vram_mb": 143771.0 |
| }, |
| { |
| "country_code": "US", |
| "cuda_max_good": 12.8, |
| "reliability": 0.9994338, |
| "total_usd_per_hour_at_listed_storage": 4.0811403508771935, |
| "usd_per_gpu_hour": 4.078947368421053, |
| "vram_mb": 143771.0 |
| }, |
| { |
| "country_code": "JP", |
| "cuda_max_good": 13.2, |
| "reliability": 0.9971439, |
| "total_usd_per_hour_at_listed_storage": 4.2911900584795335, |
| "usd_per_gpu_hour": 4.286842105263159, |
| "vram_mb": 143771.0 |
| } |
| ], |
| "eligible_offer_count": 3, |
| "query_gpu_name": "H200", |
| "quote_rule": "median of three cheapest eligible offers" |
| }, |
| "NVIDIA RTX 6000 Ada Generation": { |
| "cheapest_three_sanitized": [ |
| { |
| "country_code": "CN", |
| "cuda_max_good": 13.0, |
| "reliability": 0.9961997, |
| "total_usd_per_hour_at_listed_storage": 0.5355555555555556, |
| "usd_per_gpu_hour": 0.5333333333333333, |
| "vram_mb": 49140.0 |
| }, |
| { |
| "country_code": "JP", |
| "cuda_max_good": 13.0, |
| "reliability": 0.9939217, |
| "total_usd_per_hour_at_listed_storage": 0.5502222222222222, |
| "usd_per_gpu_hour": 0.5479999999999999, |
| "vram_mb": 49140.0 |
| }, |
| { |
| "country_code": "US", |
| "cuda_max_good": 13.0, |
| "reliability": 0.9968621, |
| "total_usd_per_hour_at_listed_storage": 0.5659259259259259, |
| "usd_per_gpu_hour": 0.56, |
| "vram_mb": 49140.0 |
| } |
| ], |
| "eligible_offer_count": 5, |
| "query_gpu_name": "RTX 6000Ada", |
| "quote_rule": "median of three cheapest eligible offers" |
| }, |
| "NVIDIA RTX A6000": { |
| "cheapest_three_sanitized": [ |
| { |
| "country_code": "RO", |
| "cuda_max_good": 13.0, |
| "reliability": 0.9916833, |
| "total_usd_per_hour_at_listed_storage": 0.3888888888888889, |
| "usd_per_gpu_hour": 0.38666666666666666, |
| "vram_mb": 49140.0 |
| } |
| ], |
| "eligible_offer_count": 1, |
| "query_gpu_name": "RTX A6000", |
| "quote_rule": "median of three cheapest eligible offers" |
| }, |
| "NVIDIA RTX PRO 6000 Blackwell Server Edition": { |
| "cheapest_three_sanitized": [ |
| { |
| "country_code": "AU", |
| "cuda_max_good": 13.2, |
| "reliability": 0.9982606, |
| "total_usd_per_hour_at_listed_storage": 1.2688888888888887, |
| "usd_per_gpu_hour": 1.2666666666666666, |
| "vram_mb": 97887.0 |
| }, |
| { |
| "country_code": "US", |
| "cuda_max_good": 13.2, |
| "reliability": 0.9967968, |
| "total_usd_per_hour_at_listed_storage": 1.2688888888888887, |
| "usd_per_gpu_hour": 1.2666666666666666, |
| "vram_mb": 97887.0 |
| }, |
| { |
| "country_code": "BG", |
| "cuda_max_good": 13.0, |
| "reliability": 0.9978267, |
| "total_usd_per_hour_at_listed_storage": 1.3362962962962963, |
| "usd_per_gpu_hour": 1.3333333333333333, |
| "vram_mb": 97887.0 |
| } |
| ], |
| "eligible_offer_count": 10, |
| "query_gpu_name": "RTX PRO 6000 S", |
| "quote_rule": "median of three cheapest eligible offers" |
| } |
| } |
| } |
|
|
