The cheapest B200 on the table right now is $1.45 per GPU-hour at Google Cloud (spot), out of 40 offers from 18 providers. The B200 carries 192 GB of VRAM on NVIDIA's Blackwell generation.
On-demand pricing for the same card spans $4.09 to $14.24 per GPU-hour — a 3.5× spread between the cheapest and the most expensive provider for identical silicon. That gap is the entire reason this table exists.
Cheapest by pricing model: on-demand $4.09 at Together AI · spot $1.45 at Google Cloud · reserved $5.49 at fal.
| Provider | Tier | $/GPU-hr | Pricing | Node | Commit | In stock | |
|---|---|---|---|---|---|---|---|
| Google Cloud | dedicated | $1.45 | spot | 1× GPU | — | — | rent |
| AWS | dedicated | $2.01 | spot | 8× GPU | — | — | rent |
| Verda | dedicated | $3.86 | spot | 1–8× GPU | — | — | rent |
| Together AI | dedicated | $4.09 | on-demand | 8× GPU | — | — | rent |
| Google Cloud | dedicated | $4.95 | on-demand | 1× GPU | — | — | rent |
| Massed Compute | dedicated | $5.43 | on-demand | 8× GPU | — | — | rent |
| fal | dedicated | $5.49 | reserved | 1× GPU | — | — | rent |
| Lium.io | marketplace | $5.50 | on-demand | 8× GPU | — | yes | rent |
| Vast.ai | marketplace | $5.94 | spot | 1× GPU | — | yes | rent |
| Hyperstack | dedicated | $6.00 | on-demand | 1× GPU | — | — | rent |
| Modal * | dedicated | $6.25 | on-demand | 1× GPU | — | yes | rent |
| Vast.ai | marketplace | $6.25 | on-demand | 1× GPU | — | yes | rent |
| Lambda Labs | dedicated | $6.69 | on-demand | 8× GPU | — | — | rent |
| Lambda Labs | dedicated | $6.79 | on-demand | 4× GPU | — | — | rent |
| Lambda Labs | dedicated | $6.89 | on-demand | 2× GPU | — | — | rent |
| Lambda Labs | dedicated | $6.99 | on-demand | 1× GPU | — | — | rent |
| Vast.ai | marketplace | $7.13 | spot | 2–8× GPU | — | yes | rent |
| Nebius * | dedicated | $7.15 | reserved | 8× GPU | — | — | rent |
| Vast.ai | marketplace | $7.50 | on-demand | 4× GPU | — | yes | rent |
| Verda | dedicated | $7.71 | on-demand | 2, 4× GPU | — | — | rent |
| Verda | dedicated | $7.72 | on-demand | 1, 8× GPU | — | — | rent |
| Runpod | dedicated | $7.99 | on-demand | 1× GPU | — | yes | rent |
| fal | dedicated | $7.99 | on-demand | 1× GPU | — | — | rent |
| GPU.ai | dedicated | $7.99 | on-demand | 1× GPU | — | yes | rent |
| Together AI | dedicated | $8.19 | reserved | 8× GPU | — | — | rent |
| Vast.ai | marketplace | $8.38 | on-demand | 8× GPU | — | yes | rent |
| Nebius * | dedicated | $8.50 | on-demand | 8× GPU | — | — | rent |
| CoreWeave | dedicated | $8.60 | on-demand | 8× GPU | — | — | rent |
| Vast.ai | marketplace | $9.38 | on-demand | 2× GPU | — | yes | rent |
| Spheron | dedicated | $9.49 | on-demand | 1× GPU | — | — | rent |
| Baseten | dedicated | $9.98 | on-demand | 1× GPU | — | yes | rent |
| Oracle Cloud | dedicated | $14.00 | on-demand | 8× GPU | — | — | rent |
| AWS | dedicated | $14.24 | on-demand | 8× GPU | — | — | rent |
How much does a B200 cost per hour?
As of the latest scrape, $1.45 per GPU-hour at Google Cloud is the cheapest B200 offer across 18 tracked cloud providers. The cheapest on-demand rate is $4.09 per GPU-hour at Together AI, and the most expensive on-demand rate is $14.24 at AWS.
Which cloud is cheapest for the B200?
Google Cloud at $1.45 per GPU-hour on spot capacity. Prices move constantly, so this page is regenerated from the live scrape every 15 minutes. Note that marketplace tiers are peer hardware and are not directly comparable to dedicated capacity.
Is spot B200 capacity cheaper?
Yes — the cheapest spot B200 is $1.45 per GPU-hour at Google Cloud, -65% versus the cheapest on-demand rate of $4.09. Spot instances can be reclaimed by the provider at any time, so they suit checkpointed training and batch inference rather than long-lived services.
How many GB of VRAM does the B200 have?
192 GB, on the Blackwell architecture. Multi-GPU instances multiply that: an 8× B200 node exposes 1536 GB of GPU memory in total.