H200 against B300 · 141 GB and 288 GB
H200 vs B300: what the price gap buys
At the lowest solid price an H200 is $3.49 per GPU-hour and a B300 is $6.60: 47% less for the H200. The H200 carries 141 GB of memory against 288 GB, so per GB of VRAM it is 8% dearer. Read 12 November 2026.
- H200 lowest
- $3.49
- B300 lowest
- $6.60
- Median gap
- -43%
Civo
Massed Compute
$5.40 vs $9.50
H200
15 providers
B300
11 providers
Price
The two price ranges, side by side
Solid on-demand rows only. Marketplace figures and announced prices are left out of both columns.
| Per GPU-hour | H200 | B300 | H200 against B300 |
|---|---|---|---|
| Providers with a price | 15 | 11 | +4 |
| Lowest | $3.49 | $6.60 | -47% |
| Median of provider lows | $5.40 | $9.50 | -43% |
| Highest | $10.60 | $17.80 | -40% |
| Highest over lowest | 3.0× | 2.7× | |
| Lowest at | Civo | Massed Compute | |
| Highest at | Google Cloud | Amazon Web Services |
The lowest H200 price is $3.11 under the lowest B300 price. 13 of the 15 H200 providers list at or under the B300 median of $9.50.
Memory
Memory, bandwidth and price per GB
VRAM is the catalogued size. Bandwidth and the maker's memory figure come from the maker's spec page, linked on each row. Nothing stated as missing is filled in.
| GPU | Variant | Memory, maker table | Type | Bandwidth | Lowest price per TB/s-hour | Source |
|---|---|---|---|---|---|---|
| H200 | SXM | 141 GB | HBM3e | 4.8 TB/s | $0.73 | page |
| H200 | NVL | 141 GB | HBM3e | 4.8 TB/s | $0.73 | page |
| B300 | SXM | 288 GB | HBM3E | 8 TB/s | $0.82 | page |
H200 · 141 GB catalogued
$0.0248
per GB-hour at the lowest price: $3.49 ÷ 141 GB
At the median: $0.0383 ($5.40 ÷ 141 GB)
B300 · 288 GB catalogued
$0.0229
per GB-hour at the lowest price: $6.60 ÷ 288 GB
At the median: $0.0330 ($9.50 ÷ 288 GB)
Price per GB-hour is a derivation: the price per GPU-hour divided by catalogued memory. It says nothing about speed, and memory is not the only reason to pick a card. The B300 carries more memory than the H200 (288 GB against 141 GB).
Fit check
Three model sizes against each card
Weights only: parameters times bytes per parameter. No KV cache, activations or framework overhead. The full table is on the memory page.
| Scenario | Weights | H200s needed | B300s needed | H200 per hour | B300 per hour |
|---|---|---|---|---|---|
| 70B model at FP16 | 140 GB | 1 | 1 | $3.49 | $6.60 |
| 70B model at FP8 | 70 GB | 1 | 1 | $3.49 | $6.60 |
| 405B model at FP8 | 405 GB | 3 | 2 | $10.47 | $13.20 |
GPUs needed = weights ÷ catalogued VRAM, rounded up. Cost = GPUs needed × the lowest solid price per GPU-hour ($3.49 for the H200, $6.60 for the B300). Real deployments need headroom and servers come in fixed sizes.
Same provider
Providers that list both
One row per provider: its lowest solid H200 price and its lowest solid B300 price. Sizes can differ, so the gap is a guide.
| Provider | H200 | B300 | Gap per GPU-hour | H200 against B300 |
|---|---|---|---|---|
| Runcrateours, live | $4.67 | $17.02 | -$12.35 | -73% |
| Amazon Web Services | $7.91 | $17.80 | -$9.89 | -56% |
| Civo | $3.49 | $7.49 | -$4.00 | -53% |
| Crusoe | $4.29 | $8.96 | -$4.67 | -52% |
| Hyperstack | $3.99 | $7.40 | -$3.41 | -46% |
| Verda | $6.50 | $12.05 | -$5.55 | -46% |
| Massed Compute | $3.62 | $6.60 | -$2.98 | -45% |
| Nebius | $5.40 | $9.50 | -$4.10 | -43% |
| RunPod | $5.29 | $8.99 | -$3.70 | -41% |
| Together AI | $5.99 | $9.99 | -$4.00 | -40% |
| Oracle Cloud | $10.00 | $15.00 | -$5.00 | -33% |
Listed for the H200 only: DigitalOcean (Paperspace), OVHcloud, CoreWeave, and Google Cloud. Every B300 provider also lists the H200.
Questions
H200 and B300, answered
- Is an H200 more expensive than a B300?
- By the lowest solid price, the H200 is 47% below the B300 ($3.49 at Civo against $6.60 at Massed Compute). By the median of each provider's lowest, $5.40 against $9.50, the H200 is 43% below.
- Which costs less per GB of memory?
- At the lowest listed price the H200 is $0.0248 per GB-hour ($3.49 ÷ 141 GB) and the B300 is $0.0229 ($6.60 ÷ 288 GB). At the medians, $0.0383 against $0.0330. VRAM is the catalogued size of each card.
- Which providers list both?
- 11 of them: Runcrate, Amazon Web Services, Civo, Crusoe, Hyperstack, Verda, Massed Compute, Nebius, RunPod, Together AI, and Oracle Cloud. At 0 of them the H200 costs more than the B300; the gap runs from -73% to -33%.
- Does the H200 need fewer GPUs for the same model?
- For weights only, a 70B model at FP16 is 140 GB: 1 H200 against 1 B300. At the lowest solid prices that is $3.49 against $6.60 per hour. This ignores the KV cache, activations and framework overhead.
- Where do the memory and bandwidth figures come from?
- Memory size is the catalogued VRAM. Bandwidth is from the maker's own spec pages, linked in the table above (www.nvidia.com and developer.nvidia.com). A figure the maker does not state stays "not stated" and gets no derived price.
Next
Other entries
Disclosure
H200 Price Ledger is published by the Runcrate team, who also run Runcrate, an AI cloud. Runcrate is listed in the table for H200, marked "ours, live". It is sorted by price like every other row, with no pinned position. Other providers' prices are read from their own public pricing pages. We have no paid placement.