01DGPU Cloud· as of 2026-10-15
Archived GPU list prices
What the clouds asked, quarter by quarter. 268 hourly list prices, 2021Q4 to 2026Q3, from 11 providers' own pricing pages and price lists as they were archived: the Wayback Machine's copies, AWS's dated price lists and Azure's effective-dated price records. List prices, not transactions, kept apart from the site's own quote snapshots and never joined to them.
B200
Each quarter shown is one where the provider’s list price changed; open it for the archived copy. Current prices are on the B200 page.
| Provider | List price, by quarter it changed | Captures |
|---|---|---|
| AWSBig cloud · p6-b200.48xlarge | 2025Q3 $14.24 | 52025Q3–2026Q3 |
| OracleBig cloud · OCI - Compute - GPU - B200 (part B110978) | 2026Q3 $14.00 | 12026Q3–2026Q3 |
| CoreWeaveSpecialist cloud · NVIDIA HGX B200 (8-GPU instance) | 2026Q2 $8.60 | 22026Q2–2026Q3 |
| HyperstackSpecialist cloud · NVIDIA B200 | 2026Q3 $6.00 | 12026Q3–2026Q3 |
| LambdaSpecialist cloud · NVIDIA B200 SXM6 (8x tab) | 2026Q1 $4.99 → 2026Q2 $6.69 | 32026Q1–2026Q3 |
| LambdaSpecialist cloud · On-demand 8x NVIDIA B200 SXM6 | 2025Q4 $4.99 | 12025Q4–2025Q4 |
| ModalSpecialist cloud · B200 | 2025Q3 $6.25Modal's serverless per-second GPU price, times 3,600: not comparable with instance prices. | 12025Q3–2025Q3 |
| ModalSpecialist cloud · Nvidia B200 | 2025Q4 $6.25Modal's serverless per-second GPU price, times 3,600: not comparable with instance prices. | 42025Q4–2026Q3 |
| NebiusSpecialist cloud · NVIDIA HGX B200 | 2025Q3 $5.50 → 2026Q3 $7.15 | 52025Q3–2026Q3 |
| Together AISpecialist cloud · NVIDIA HGX B200 | 2025Q4 $5.50 → 2026Q2 $7.49 → 2026Q3 $8.19 | 42025Q4–2026Q3 |
H200
Each quarter shown is one where the provider’s list price changed; open it for the archived copy. Current prices are on the H200 page.
| Provider | List price, by quarter it changed | Captures |
|---|---|---|
| AWSBig cloud · p5en.48xlarge | 2025Q1 $10.60 → 2025Q3 $7.91 | 72025Q1–2026Q3 |
| AzureBig cloud · Standard_ND96isr_H200_v5 | 2025Q2 $10.60 | 1effective from 2025Q2 |
| Google CloudBig cloud · a3-ultragpu-8g | 2025Q2 $10.60 | 62025Q2–2026Q3 |
| OracleBig cloud · OCI - Compute - GPU - H200 (part B110519) | 2026Q3 $10.00 | 12026Q3–2026Q3 |
| CoreWeaveSpecialist cloud · NVIDIA HGX H200 (8-GPU instance) | 2025Q1 $6.30 | 72025Q1–2026Q3 |
| CrusoeSpecialist cloud · NVIDIA H200 141GB HGX | 2025Q4 $4.29 | 42025Q4–2026Q3 |
| CrusoeSpecialist cloud · NVIDIA H200 141GB SXM | 2025Q2 $4.29 | 22025Q2–2025Q3 |
| HyperstackSpecialist cloud · NVIDIA H200 SXM | 2025Q3 $3.50 → 2026Q3 $3.99 | 52025Q3–2026Q3 |
| ModalSpecialist cloud · H200 | 2025Q3 $4.54Modal's serverless per-second GPU price, times 3,600: not comparable with instance prices. | 12025Q3–2025Q3 |
| ModalSpecialist cloud · Nvidia H200 | 2025Q4 $4.54Modal's serverless per-second GPU price, times 3,600: not comparable with instance prices. | 42025Q4–2026Q3 |
| NebiusSpecialist cloud · 8 × H200 SXM | 2024Q4 $3.50 | 12024Q4–2024Q4 |
| NebiusSpecialist cloud · NVIDIA H200 GPU | 2025Q1 $3.50 | 22025Q1–2025Q2 |
| NebiusSpecialist cloud · NVIDIA HGX H200 | 2025Q3 $3.50 → 2026Q3 $4.50 | 52025Q3–2026Q3 |
| Together AISpecialist cloud · NVIDIA HGX H200 | 2025Q4 $3.79 → 2026Q2 $4.19 → 2026Q3 $5.99 | 42025Q4–2026Q3 |
H100
Each quarter shown is one where the provider’s list price changed; open it for the archived copy. Current prices are on the H100 page.
| Provider | List price, by quarter it changed | Captures |
|---|---|---|
| AWSBig cloud · p5.48xlarge | 2023Q3 $12.29 → 2025Q3 $6.88 | 132023Q3–2026Q3 |
| AzureBig cloud · Standard_ND96isr_H100_v5 | 2023Q4 $12.29 | 1effective from 2023Q4 |
| Google CloudBig cloud · a3-highgpu-8g | 2024Q2 $11.06 | 102024Q2–2026Q3 |
| Google CloudBig cloud · a3-megagpu-8g | 2025Q2 $11.68 | 62025Q2–2026Q3 |
| OracleBig cloud · Oracle Cloud Infrastructure - Compute - GPU - H100 (part B98415) = BM.GPU.H100.8 (8x NVIDIA H100 80GB) | 2026Q3 $10.00 | 12026Q3–2026Q3 |
| CoreWeaveSpecialist cloud · NVIDIA HGX H100 | 2023Q2 $4.76CoreWeave's GPU-only component price, before CPU and memory: not comparable with full-instance prices. | 72023Q2–2024Q4 |
| CoreWeaveSpecialist cloud · NVIDIA HGX H100 (8-GPU instance) | 2025Q1 $6.16 | 72025Q1–2026Q3 |
| CrusoeSpecialist cloud · NVIDIA H100 80GB HGX | 2025Q4 $3.90 | 42025Q4–2026Q3 |
| CrusoeSpecialist cloud · NVIDIA H100 80GB SXM | 2025Q2 $3.90 | 22025Q2–2025Q3 |
| HyperstackSpecialist cloud · NVIDIA H100 SXM | 2025Q3 $2.40 → 2026Q3 $3.20 | 52025Q3–2026Q3 |
| HyperstackSpecialist cloud · NVIDIA H100 SXM 80GB | 2024Q3 $3.75 → 2024Q4 $3.00 → 2025Q2 $2.40 | 42024Q3–2025Q2 |
| LambdaSpecialist cloud · 8x NVIDIA H100 SXM New | 2023Q3 $2.59 → 2024Q1 $3.49 | 42023Q3–2024Q2 |
| LambdaSpecialist cloud · 8x NVIDIA H100 SXM New Price | 2024Q3 $2.99 | 12024Q3–2024Q3 |
| LambdaSpecialist cloud · NVIDIA H100 SXM (8x tab) | 2026Q1 $2.99 → 2026Q2 $3.99 | 32026Q1–2026Q3 |
| LambdaSpecialist cloud · On-demand 8x NVIDIA H100 SXM | 2024Q4 $2.99 | 52024Q4–2025Q4 |
| ModalSpecialist cloud · H100 | 2025Q3 $3.95Modal's serverless per-second GPU price, times 3,600: not comparable with instance prices. | 12025Q3–2025Q3 |
| ModalSpecialist cloud · Nvidia H100 | 2024Q1 $7.65 → 2024Q3 $5.92 → 2024Q4 $4.56 → 2025Q1 $3.95Modal's serverless per-second GPU price, times 3,600: not comparable with instance prices. | 102024Q1–2026Q3 |
| NebiusSpecialist cloud · 8 × H100 SXM | 2024Q4 $2.95 | 12024Q4–2024Q4 |
| NebiusSpecialist cloud · H100 SXM5 | 2023Q4 $4.85 | 42023Q4–2024Q3 |
| NebiusSpecialist cloud · NVIDIA H100 GPU | 2025Q1 $2.95 | 22025Q1–2025Q2 |
| NebiusSpecialist cloud · NVIDIA HGX H100 | 2025Q3 $2.95 → 2026Q3 $3.85 | 52025Q3–2026Q3 |
| Together AISpecialist cloud · NVIDIA HGX H100 | 2026Q2 $3.49 → 2026Q3 $3.99 | 22026Q2–2026Q3 |
| Together AISpecialist cloud · NVIDIA HGX H100 SXM | 2025Q4 $2.99 | 22025Q4–2026Q1 |
A100 80GB
Each quarter shown is one where the provider’s list price changed; open it for the archived copy. Current prices are on the A100 80GB page.
| Provider | List price, by quarter it changed | Captures |
|---|---|---|
| AWSBig cloud · p4de.24xlarge | 2022Q2 $5.12 → 2025Q3 $3.43 | 182022Q2–2026Q3 |
| AzureBig cloud · Standard_ND96amsr_A100_v4 | 2021Q4 $4.10 | 1effective from 2021Q4 |
| Google CloudBig cloud · a2-ultragpu-8g | 2023Q1 $5.07 | 152023Q1–2026Q3 |
| OracleBig cloud · Compute - GPU - A100 - v2 (part B95907) = BM.GPU.A100-v2.8 (8x NVIDIA A100 80GB) | 2026Q3 $4.00 | 12026Q3–2026Q3 |
| CoreWeaveSpecialist cloud · A100 80GB NVLINK | 2022Q3 $2.21CoreWeave's GPU-only component price, before CPU and memory: not comparable with full-instance prices. | 102022Q3–2024Q4 |
| CoreWeaveSpecialist cloud · NVIDIA A100 (8-GPU instance) | 2025Q2 $2.70 | 62025Q2–2026Q3 |
| CrusoeSpecialist cloud · NVIDIA A100 80GB SXM | 2025Q2 $1.95 → 2026Q2 $2.30 | 62025Q2–2026Q3 |
| CrusoeSpecialist cloud · NVIDIA A100 Tensor Core GPU 80GB SXM | 2024Q2 $1.95 | 22024Q2–2024Q4 |
| HyperstackSpecialist cloud · NVIDIA A100 SXM | 2025Q3 $1.60 | 52025Q3–2026Q3 |
| LambdaSpecialist cloud · 8x NVIDIA A100 | 2023Q1 $1.50 | 42023Q1–2023Q4 |
| LambdaSpecialist cloud · 8x NVIDIA A100 SXM | 2024Q1 $1.79 | 32024Q1–2024Q3 |
| LambdaSpecialist cloud · NVIDIA A100 SXM (8x tab) | 2026Q1 $1.79 → 2026Q2 $2.79 | 32026Q1–2026Q3 |
| LambdaSpecialist cloud · On-demand 8x NVIDIA A100 SXM | 2024Q4 $1.79 | 52024Q4–2025Q4 |
| ModalSpecialist cloud · A100-80GB | 2025Q3 $2.50Modal's serverless per-second GPU price, times 3,600: not comparable with instance prices. | 12025Q3–2025Q3 |
| ModalSpecialist cloud · Nvidia A100, 80 GB | 2023Q4 $5.59 → 2024Q3 $4.75 → 2024Q4 $3.40 → 2025Q1 $2.50Modal's serverless per-second GPU price, times 3,600: not comparable with instance prices. | 112023Q4–2026Q3 |
Background reading: Cloud AI GPU Pricing Comparison on AI Wiki. AI Wiki is run by the same team as AI Compute Tracker.