# PetaPrice

> Compare live GPU cloud prices from 35+ providers, normalized to dollars per PetaFLOP-hour. Find the cheapest H100, H200, B200, MI300X, RTX 4090 and more by VRAM, precision and price.

Data as of 2026-09-27T01:12:14.555Z (refreshed about every 2 hours): 958 offers from 39 providers.

## Definitions
- Price: USD per GPU per hour, list price, excluding taxes, storage and egress unless bundled.
- $/PFLOP-hour: price per GPU-hour divided by dense peak PFLOPS at a given precision (default FP16/BF16). Lower is better.
- Dense FLOPS: vendor datasheet peaks without 2:4 sparsity. GeForce cards use the FP32-accumulate tensor rate.
- On-demand, spot (interruptible) and reserved (committed) prices are reported separately.
- Monthly cost: hourly price × 730.

## Best compute per dollar (FP16/BF16, on-demand)
1. V100 SXM2 16GB: $0.65/PFLOP-hour, from $0.08/GPU-hour (10 providers) — https://petaprice.com/gpu/v100-sxm-16
2. Instinct MI355X: $1.03/PFLOP-hour, from $2.59/GPU-hour (3 providers) — https://petaprice.com/gpu/mi355x
3. RTX A4000: $1.04/PFLOP-hour, from $0.08/GPU-hour (5 providers) — https://petaprice.com/gpu/rtx-a4000
4. RTX PRO 6000 Blackwell: $1.17/PFLOP-hour, from $0.59/GPU-hour (16 providers) — https://petaprice.com/gpu/rtx-pro-6000
5. V100 SXM2 32GB: $1.19/PFLOP-hour, from $0.15/GPU-hour (2 providers) — https://petaprice.com/gpu/v100-sxm-32
6. Instinct MI300X: $1.31/PFLOP-hour, from $1.71/GPU-hour (9 providers) — https://petaprice.com/gpu/mi300x
7. RTX 5070 Ti: $1.32/PFLOP-hour, from $0.12/GPU-hour (1 provider) — https://petaprice.com/gpu/rtx-5070-ti
8. RTX A5000: $1.44/PFLOP-hour, from $0.16/GPU-hour (4 providers) — https://petaprice.com/gpu/rtx-a5000
9. RTX 4070 SUPER: $1.54/PFLOP-hour, from $0.11/GPU-hour (1 provider) — https://petaprice.com/gpu/rtx-4070-super
10. RTX 6000 Ada: $1.65/PFLOP-hour, from $0.60/GPU-hour (6 providers) — https://petaprice.com/gpu/rtx-6000-ada
11. B200 SXM: $1.66/PFLOP-hour, from $3.74/GPU-hour (14 providers) — https://petaprice.com/gpu/b200
12. L40S: $1.66/PFLOP-hour, from $0.60/GPU-hour (17 providers) — https://petaprice.com/gpu/l40s
13. H100 SXM: $1.68/PFLOP-hour, from $1.66/GPU-hour (27 providers) — https://petaprice.com/gpu/h100-sxm
14. V100 PCIe: $1.70/PFLOP-hour, from $0.19/GPU-hour (2 providers) — https://petaprice.com/gpu/v100-pcie
15. A100 SXM 40GB: $1.71/PFLOP-hour, from $0.54/GPU-hour (7 providers) — https://petaprice.com/gpu/a100-sxm-40

## GPU pages
- [GB300](https://petaprice.com/gpu/gb300): from $9.62/GPU-hour, median $13.8, 288 GB
- [B300 SXM](https://petaprice.com/gpu/b300): from $4.50/GPU-hour, median $8.37, 288 GB
- [GB200](https://petaprice.com/gpu/gb200): from $16.0/GPU-hour, median $21.5, 186 GB
- [B200 SXM](https://petaprice.com/gpu/b200): from $3.74/GPU-hour, median $6.80, 180 GB
- [H200 SXM](https://petaprice.com/gpu/h200-sxm): from $2.10/GPU-hour, median $3.99, 141 GB
- [H200 NVL](https://petaprice.com/gpu/h200-nvl): from $3.29/GPU-hour, median $3.54, 141 GB
- [GH200](https://petaprice.com/gpu/gh200): from $1.99/GPU-hour, median $2.29, 96 GB
- [H100 SXM](https://petaprice.com/gpu/h100-sxm): from $1.66/GPU-hour, median $3.19, 80 GB
- [H100 NVL](https://petaprice.com/gpu/h100-nvl): from $2.59/GPU-hour, median $2.92, 94 GB
- [H100 PCIe](https://petaprice.com/gpu/h100-pcie): from $1.98/GPU-hour, median $2.75, 80 GB
- [A100 SXM 80GB](https://petaprice.com/gpu/a100-sxm-80): from $0.81/GPU-hour, median $2.35, 80 GB
- [A100 SXM 40GB](https://petaprice.com/gpu/a100-sxm-40): from $0.54/GPU-hour, median $1.99, 40 GB
- [A100 PCIe 80GB](https://petaprice.com/gpu/a100-pcie-80): from $1.13/GPU-hour, median $1.51, 80 GB
- [A100 PCIe 40GB](https://petaprice.com/gpu/a100-pcie-40): from $1.99/GPU-hour, median $1.99, 40 GB
- [L40S](https://petaprice.com/gpu/l40s): from $0.60/GPU-hour, median $1.50, 48 GB
- [L40](https://petaprice.com/gpu/l40): from $0.69/GPU-hour, median $0.88, 48 GB
- [L4](https://petaprice.com/gpu/l4): from $0.27/GPU-hour, median $0.56, 24 GB
- [A40](https://petaprice.com/gpu/a40): from $0.49/GPU-hour, median $0.49, 48 GB
- [A30](https://petaprice.com/gpu/a30): from $0.29/GPU-hour, median $0.33, 24 GB
- [A10](https://petaprice.com/gpu/a10): from $1.00/GPU-hour, median $1.65, 24 GB
- [A10G](https://petaprice.com/gpu/a10g): from $1.01/GPU-hour, median $1.01, 24 GB
- [A16](https://petaprice.com/gpu/a16): from $0.47/GPU-hour, median $0.47, 16 GB
- [T4](https://petaprice.com/gpu/t4): from $0.34/GPU-hour, median $0.35, 16 GB
- [V100 SXM2 32GB](https://petaprice.com/gpu/v100-sxm-32): from $0.15/GPU-hour, median $2.03, 32 GB
- [V100 SXM2 16GB](https://petaprice.com/gpu/v100-sxm-16): from $0.08/GPU-hour, median $1.57, 16 GB
- [V100 PCIe](https://petaprice.com/gpu/v100-pcie): from $0.19/GPU-hour, median $1.63, 16 GB
- [V100S PCIe](https://petaprice.com/gpu/v100s-pcie): from $0.88/GPU-hour, median $0.88, 32 GB
- [P100](https://petaprice.com/gpu/p100): from $1.28/GPU-hour, median $1.33, 16 GB
- [RTX PRO 6000 Blackwell](https://petaprice.com/gpu/rtx-pro-6000): from $0.59/GPU-hour, median $1.56, 96 GB
- [RTX 6000 Ada](https://petaprice.com/gpu/rtx-6000-ada): from $0.60/GPU-hour, median $0.86, 48 GB
- [RTX 5000 Ada](https://petaprice.com/gpu/rtx-5000-ada): from $0.49/GPU-hour, median $0.49, 32 GB
- [RTX 4000 Ada](https://petaprice.com/gpu/rtx-4000-ada): from $0.20/GPU-hour, median $0.48, 20 GB
- [RTX A6000](https://petaprice.com/gpu/rtx-a6000): from $0.33/GPU-hour, median $0.56, 48 GB
- [RTX A5000](https://petaprice.com/gpu/rtx-a5000): from $0.16/GPU-hour, median $0.34, 24 GB
- [RTX A4500](https://petaprice.com/gpu/rtx-a4500): from $0.19/GPU-hour, median $0.19, 20 GB
- [RTX A4000](https://petaprice.com/gpu/rtx-a4000): from $0.08/GPU-hour, median $0.15, 16 GB
- [RTX PRO 5000 Blackwell](https://petaprice.com/gpu/rtx-pro-5000): from $0.82/GPU-hour, median $0.82, 48 GB
- [RTX PRO 4000 Blackwell](https://petaprice.com/gpu/rtx-pro-4000): from $0.57/GPU-hour, median $0.57, 24 GB
- [RTX 5880 Ada](https://petaprice.com/gpu/rtx-5880-ada): from $0.67/GPU-hour, median $0.67, 48 GB
- [RTX PRO 4500 Blackwell](https://petaprice.com/gpu/rtx-pro-4500): from $0.72/GPU-hour, median $0.72, 32 GB
- [RTX 4000 SFF Ada](https://petaprice.com/gpu/rtx-4000-sff-ada): from $0.18/GPU-hour, median $0.18, 20 GB
- [Quadro RTX 6000](https://petaprice.com/gpu/quadro-rtx-6000): from $0.33/GPU-hour, median $0.51, 24 GB
- [RTX 2000 Ada](https://petaprice.com/gpu/rtx-2000-ada): from $0.24/GPU-hour, median $0.24, 16 GB
- [RTX A2000](https://petaprice.com/gpu/rtx-a2000): from $0.12/GPU-hour, median $0.12, 12 GB
- [RTX 5090](https://petaprice.com/gpu/rtx-5090): from $0.40/GPU-hour, median $0.65, 32 GB
- [RTX 5080](https://petaprice.com/gpu/rtx-5080): from $0.25/GPU-hour, median $0.27, 16 GB
- [RTX 5070 Ti](https://petaprice.com/gpu/rtx-5070-ti): from $0.12/GPU-hour, median $0.12, 16 GB
- [RTX 4090](https://petaprice.com/gpu/rtx-4090): from $0.29/GPU-hour, median $0.39, 24 GB
- [RTX 4080 SUPER](https://petaprice.com/gpu/rtx-4080-super): from $0.27/GPU-hour, median $0.27, 16 GB
- [RTX 4080](https://petaprice.com/gpu/rtx-4080): from $0.23/GPU-hour, median $0.25, 16 GB
- [RTX 4070 Ti](https://petaprice.com/gpu/rtx-4070-ti): from $0.19/GPU-hour, median $0.22, 12 GB
- [RTX 4070](https://petaprice.com/gpu/rtx-4070): from $0.14/GPU-hour, median $0.14, 12 GB
- [RTX 4060 Ti](https://petaprice.com/gpu/rtx-4060-ti): from $0.09/GPU-hour, median $0.09, 16 GB
- [RTX 3090 Ti](https://petaprice.com/gpu/rtx-3090-ti): from $0.27/GPU-hour, median $0.27, 24 GB
- [RTX 3090](https://petaprice.com/gpu/rtx-3090): from $0.13/GPU-hour, median $0.17, 24 GB
- [RTX 3080 Ti](https://petaprice.com/gpu/rtx-3080-ti): from $0.18/GPU-hour, median $0.18, 12 GB
- [RTX 3080](https://petaprice.com/gpu/rtx-3080): from $0.13/GPU-hour, median $0.14, 10 GB
- [RTX 3070](https://petaprice.com/gpu/rtx-3070): from $0.13/GPU-hour, median $0.13, 8 GB
- [RTX 5060 Ti](https://petaprice.com/gpu/rtx-5060-ti): from $0.11/GPU-hour, median $0.11, 16 GB
- [RTX 4070 SUPER](https://petaprice.com/gpu/rtx-4070-super): from $0.11/GPU-hour, median $0.11, 12 GB
- [Instinct MI355X](https://petaprice.com/gpu/mi355x): from $2.59/GPU-hour, median $2.85, 288 GB
- [Instinct MI350X](https://petaprice.com/gpu/mi350x): from $6.16/GPU-hour, median $6.16, 288 GB
- [Instinct MI325X](https://petaprice.com/gpu/mi325x): from $2.25/GPU-hour, median $3.80, 256 GB
- [Instinct MI300X](https://petaprice.com/gpu/mi300x): from $1.71/GPU-hour, median $2.99, 192 GB
- [Gaudi 2](https://petaprice.com/gpu/gaudi2): from $1.10/GPU-hour, median $1.10, 96 GB
- [TPU v6e (Trillium)](https://petaprice.com/gpu/tpu-v6e): from $2.70/GPU-hour, median $2.70, 32 GB
- [TPU v5p](https://petaprice.com/gpu/tpu-v5p): from $4.20/GPU-hour, median $4.20, 95 GB
- [TPU v5e](https://petaprice.com/gpu/tpu-v5e): from $1.20/GPU-hour, median $1.20, 16 GB
- [Trainium](https://petaprice.com/gpu/trn1): from $1.34/GPU-hour, median $1.34, 32 GB
- [Inferentia2](https://petaprice.com/gpu/inf2): from $0.76/GPU-hour, median $0.76, 32 GB

## Comparisons
- https://petaprice.com/compare/h100-sxm-vs-h200-sxm
- https://petaprice.com/compare/h100-sxm-vs-a100-sxm-80
- https://petaprice.com/compare/h100-sxm-vs-b200
- https://petaprice.com/compare/h200-sxm-vs-b200
- https://petaprice.com/compare/b200-vs-b300
- https://petaprice.com/compare/gb200-vs-b200
- https://petaprice.com/compare/h100-sxm-vs-mi300x
- https://petaprice.com/compare/h200-sxm-vs-mi300x
- https://petaprice.com/compare/mi300x-vs-mi325x
- https://petaprice.com/compare/b200-vs-mi355x
- https://petaprice.com/compare/h100-pcie-vs-h100-sxm
- https://petaprice.com/compare/h100-nvl-vs-h100-sxm
- https://petaprice.com/compare/gh200-vs-h100-sxm
- https://petaprice.com/compare/a100-sxm-80-vs-a100-pcie-80
- https://petaprice.com/compare/a100-sxm-40-vs-a100-sxm-80
- https://petaprice.com/compare/l40s-vs-a100-sxm-80
- https://petaprice.com/compare/l40s-vs-h100-pcie
- https://petaprice.com/compare/l40s-vs-rtx-6000-ada
- https://petaprice.com/compare/rtx-pro-6000-vs-l40s
- https://petaprice.com/compare/rtx-pro-6000-vs-h100-sxm
- https://petaprice.com/compare/rtx-4090-vs-rtx-5090
- https://petaprice.com/compare/rtx-5090-vs-rtx-pro-6000
- https://petaprice.com/compare/rtx-4090-vs-a100-sxm-80
- https://petaprice.com/compare/rtx-3090-vs-rtx-4090
- https://petaprice.com/compare/rtx-a6000-vs-rtx-6000-ada
- https://petaprice.com/compare/a10-vs-l4
- https://petaprice.com/compare/t4-vs-l4
- https://petaprice.com/compare/v100-sxm-32-vs-a100-sxm-40

## API
- https://petaprice.com/api/offers — every normalized offer plus the GPU spec sheet (JSON, CORS enabled).

## Other
- Methodology: https://petaprice.com/methodology
- Imprint: https://petaprice.com/imprint

## Citation
Cite as: "PetaPrice, <page title>, <URL>, accessed 2026-09-27". Prices are indicative and change frequently.