Amazon Web Services vs Vast.ai
As of , Amazon Web Services and Vast.ai share 7 GPUs today, but on none do both list an on-demand rate: at least one side is community capacity or a marketplace floor, compared below and not counted. Not compared today: 2 more shared GPUs, where one side's quote is not confirmed within 48 hours or is a thin market. On the priciest shared card, NVIDIA B200: Amazon Web Services $14.24 (us-east-1) vs Vast.ai $10.25 (Multi-GPU machines · floor) per GPU-hour — not the same market, not counted.
On-demand, per GPU-hour
| GPU |  Amazon Web Services |  Vast.ai | Difference | Cheaper |
|---|---|---|---|---|
| NVIDIA B200 median $6.99 · 35 providers · lowest Beam $4.11 from | $14.24 us-east-1 captured Deploy ↗ | $10.25 Multi-GPU machines · floor captured Deploy ↗ | — | not the same market |
| NVIDIA A100 80GB median $2.00 · 58 providers · lowest CloudRift $1.05 | $3.43 us-east-1 captured Deploy ↗ | $1.07 Multi-GPU machines · floor captured Deploy ↗ | — | not the same market |
| NVIDIA RTX PRO 6000 Blackwell median $2.21 · 38 providers · lowest Hostinger $0.65 from | $3.36 us-east-1 captured Deploy ↗ | $1.34 floor captured Deploy ↗ | — | not the same market |
| NVIDIA V100 median $2.48 · 11 providers · lowest Verda $0.23 | $3.06 us-east-1 captured Deploy ↗ | $0.11 floor captured Deploy ↗ | — | not the same market |
| NVIDIA A100 40GB median $1.98 · 20 providers · lowest Trooper.AI $0.68 from | $2.74 us-east-1 captured Deploy ↗ | $0.42 floor captured Deploy ↗ | — | not the same market |
| NVIDIA L40S median $1.40 · 38 providers · lowest CloudRift $0.63 | $1.86 us-east-1 captured Deploy ↗ | $0.79 floor captured Deploy ↗ | — | not the same market |
| NVIDIA L4 median $0.80 · 21 providers · lowest Hikube $0.27 GPU only | $0.80 us-east-1 captured Deploy ↗ | $0.36 floor captured Deploy ↗ | — | not the same market |
Shared, not compared today
Both providers quoted these GPUs on-demand within the last 7 days, but at least one quote is not confirmed within 48 hours of the newest read (or is a marketplace floor withheld as too thin), and an unconfirmed price never decides which is cheaper. They return to the table above at the next read that confirms them.
- NVIDIA H100 SXM: Vast.ai: thin market
- NVIDIA H200: Vast.ai: thin market
How this is compared
Cloud GPU rental prices of Amazon Web Services and Vast.ai, compared per GPU-hour on the GPUs both list. Where a provider has several tiers or regions, its cheapest on-demand rate is used and labelled. Community capacity (RunPod Community Cloud) and marketplace floors are shown only where a provider has nothing else, and never decide which is cheaper.
GPU by GPU
- NVIDIA B200: Amazon Web Services $14.24 (us-east-1) vs Vast.ai $10.25 (Multi-GPU machines · floor) per GPU-hour — not the same market, not counted.
- NVIDIA A100 80GB: Amazon Web Services $3.43 (us-east-1) vs Vast.ai $1.07 (Multi-GPU machines · floor) per GPU-hour — not the same market, not counted.
- NVIDIA RTX PRO 6000 Blackwell: Amazon Web Services $3.36 (us-east-1) vs Vast.ai $1.34 (floor) per GPU-hour — not the same market, not counted.
- NVIDIA V100: Amazon Web Services $3.06 (us-east-1) vs Vast.ai $0.11 (floor) per GPU-hour — not the same market, not counted.
- NVIDIA A100 40GB: Amazon Web Services $2.74 (us-east-1) vs Vast.ai $0.42 (floor) per GPU-hour — not the same market, not counted.
- NVIDIA L40S: Amazon Web Services $1.86 (us-east-1) vs Vast.ai $0.79 (floor) per GPU-hour — not the same market, not counted.
- NVIDIA L4: Amazon Web Services $0.80 (us-east-1) vs Vast.ai $0.36 (floor) per GPU-hour — not the same market, not counted.
Not compared: GPUs only one of the two lists, reserved/committed tiers, and anything either provider prices only on request. See each provider's page for its full list.
Data updated: