H100 vs B200 rental price per hour
The B200 costs 100% more per hour than the H100 at the median and has 125% more memory (180 GB against 80 GB). Per GB of memory, the B200 is 11% cheaper. Based on 28 H100 and 13 B200 public on-demand listings, last checked .
- H100 median
- $3.99
- 17 providers
- B200 median
- $7.99
- 10 providers
- H100 memory
- 80 GB
- Per GPU
- B200 memory
- 180 GB
- Per GPU
Side by side
H100 and B200 public on-demand prices
US dollars per GPU-hour from public pages. Every row behind these numbers is on the H100 price page and the B200 price page, each with its source link.
| Measure | H100 | B200 |
|---|---|---|
| Memory per GPU | 80 GB | 180 GB |
| Cheapest on-demand | $1.68 (Latitude.sh) | $5.43 (Massed Compute) |
| Median on-demand | $3.99 | $7.99 |
| Priciest on-demand | $12.29 | $14.24 |
| Providers with a price | 17 | 10 |
| Median price per GB of memory, per hour | $0.050 | $0.044 |
Same provider
Providers that list both
Lowest public on-demand price per GPU-hour at each provider that lists both chips.
| Provider | H100 lowest | B200 lowest | B200 vs H100 |
|---|---|---|---|
| Amazon Web Services | $6.88 | $14.24 | +107% |
| CoreWeave | $6.16 | $8.60 | +40% |
| Hyperstack | $2.50 | $6.00 | +140% |
| Lambda | $3.79 | $7.69 | +103% |
| Massed Compute | $2.73 | $5.43 | +99% |
| Nebius | $4.50 | $8.50 | +89% |
| Oracle Cloud (OCI) | $10.00 | $14.00 | +40% |
| RunPod | $2.89 | $7.99 | +176% |
| Together AI | $3.99 | $8.19 | +105% |
| Verda | $4.99 | $9.14 | +83% |
FAQ
H100 vs B200 questions
Is the B200 more expensive than the H100 to rent?
At the median public on-demand price, yes by 100%: $7.99 against $3.99 per GPU-hour.
Which has more memory, the H100 or the B200?
The H100 has 80 GB and the B200 has 180 GB per GPU, as published by the vendors and shown in provider listings.
Where is the cheapest H100 and B200?
The cheapest public on-demand H100 we track is $1.68 at Latitude.sh. For the B200 it is $5.43 at Massed Compute. Last checked Nov 9, 2026.
Need H100 or B200 capacity?
Tell us the GPU, count, term, region and budget. We reply with the public prices that match and the checks to run before you pay. Free, and we do not sell compute.
More