# Price of Compute — full machine reference

> GPU rental prices (USD per GPU-hour, cross-provider daily medians) and verified LLM token prices, with a composite index. Free API, attribution required: "Data: Price of Compute — priceofcompute.com". Method: per-provider medians, then median across providers; on-demand/spot/community never mixed; gaps never interpolated. Archival history (labeled) back to January 2023.

Generated: 2026-08-11T04:37:37.153Z

## The Index
- Official daily close: 119.6 (2026-08-11, base 100 = 2026-08-10)
- Live print: 119.62 as of 2026-08-11T03:30:49.661Z (5/5 basket SKUs live)

## Today's on-demand medians (USD per GPU-hour)
- NVIDIA B300: $6.70 (https://priceofcompute.com/gpus/b300)
- NVIDIA B200: $6.51 (https://priceofcompute.com/gpus/b200)
- NVIDIA H200 SXM: $4.59 (https://priceofcompute.com/gpus/h200-sxm)
- NVIDIA H100 SXM: $3.87 (https://priceofcompute.com/gpus/h100-sxm)
- NVIDIA H200 NVL: $3.72 (https://priceofcompute.com/gpus/h200-nvl)
- NVIDIA H100 NVL: $3.10 (https://priceofcompute.com/gpus/h100-nvl)
- AMD Instinct MI300X: $2.69 (https://priceofcompute.com/gpus/mi300x)
- NVIDIA H100 PCIe: $2.60 (https://priceofcompute.com/gpus/h100-pcie)
- NVIDIA RTX PRO 6000 Blackwell: $2.19 (https://priceofcompute.com/gpus/rtx-pro-6000)
- NVIDIA A100 80GB SXM: $1.82 (https://priceofcompute.com/gpus/a100-sxm-80gb)
- NVIDIA L40S: $1.52 (https://priceofcompute.com/gpus/l40s)
- NVIDIA A100 40GB SXM: $1.45 (https://priceofcompute.com/gpus/a100-sxm-40gb)
- NVIDIA A100 80GB PCIe: $1.40 (https://priceofcompute.com/gpus/a100-pcie-80gb)
- NVIDIA A100 40GB PCIe: $1.15 (https://priceofcompute.com/gpus/a100-pcie-40gb)
- NVIDIA RTX 5090: $0.99 (https://priceofcompute.com/gpus/rtx-5090)
- NVIDIA L40: $0.82 (https://priceofcompute.com/gpus/l40)
- NVIDIA RTX A6000: $0.79 (https://priceofcompute.com/gpus/rtx-a6000)
- NVIDIA RTX 6000 Ada: $0.77 (https://priceofcompute.com/gpus/rtx-6000-ada)
- NVIDIA RTX 4090: $0.74 (https://priceofcompute.com/gpus/rtx-4090)
- NVIDIA RTX 3090: $0.34 (https://priceofcompute.com/gpus/rtx-3090)

## Tracked providers
- Azure: 5 SKUs tracked
- CoreWeave: 7 SKUs tracked
- DataCrunch: 8 SKUs tracked
- Denvr: 4 SKUs tracked
- Hot Aisle: 1 SKUs tracked
- Hyperstack: 8 SKUs tracked
- Lambda: 7 SKUs tracked
- Massed Compute: 9 SKUs tracked
- Nebius: 4 SKUs tracked
- OVHcloud: 1 SKUs tracked
- RunPod: 19 SKUs tracked
- SaladCloud: 2 SKUs tracked
- Thunder Compute: 2 SKUs tracked
- Vast.ai: 19 SKUs tracked
- Voltage Park: 1 SKUs tracked
- Vultr: 1 SKUs tracked

## LLM token prices (USD per million tokens, standard tier, verified)
- anthropic/claude-fable-5: input $10.00, output $50.00 (verified 2026-06-24)
- anthropic/claude-opus-4-8: input $5.00, output $25.00 (verified 2026-06-24)
- anthropic/claude-opus-5: input $5.00, output $25.00 (verified 2026-06-24)
- anthropic/claude-sonnet-5: input $2.00, output $10.00 (verified 2026-06-24)
- anthropic/claude-haiku-4-5: input $1.00, output $5.00 (verified 2026-06-24)
- deepseek/deepseek-v4-pro: input $0.43, output $0.87 (verified 2026-08-11)
- deepseek/deepseek-v4-flash: input $0.14, output $0.28 (verified 2026-08-11)
- google/gemini-3.1-pro-preview: input $2.00, output $12.00 (verified 2026-08-10)
- google/gemini-3.5-flash: input $1.50, output $9.00 (verified 2026-08-10)
- google/gemini-3.6-flash: input $1.50, output $7.50 (verified 2026-08-10)
- google/gemini-3.5-flash-lite: input $0.30, output $2.50 (verified 2026-08-10)
- mistral/mistral-medium-3.5: input $1.50, output $7.50 (verified 2026-08-11)
- mistral/mistral-large-3: input $0.50, output $1.50 (verified 2026-08-11)
- mistral/mistral-small-4: input $0.15, output $0.60 (verified 2026-08-11)
- openai/gpt-5.5-pro: input $30.00, output $180.00 (verified 2026-08-10)
- openai/gpt-5.6-sol: input $5.00, output $30.00 (verified 2026-08-10)
- openai/gpt-5.4: input $2.50, output $15.00 (verified 2026-08-10)
- openai/gpt-5.6-terra: input $2.00, output $12.00 (verified 2026-08-10)
- openai/gpt-5.6-luna: input $0.20, output $1.20 (verified 2026-08-10)
- xai/grok-4.5: input $2.00, output $6.00 (verified 2026-08-11)

## API
- Base: https://priceofcompute.com/api/v1 — endpoints listed at /api/v1/openapi.json
- 1,000 requests/day per IP. Attribution required.

## Citation
- "Price of Compute, priceofcompute.com, accessed <date>." Details: https://priceofcompute.com/cite