B300 SXM 288GB vs H100 SXM5 80GB — current cloud rental cost per hour, normalized to USD per GPU across providers. H100 SXM5 80GB is currently the cheaper option — about 47% less on the headline comparison metric.
| Context | B300 SXM 288GB | H100 SXM5 80GB |
|---|---|---|
| Architecture | NVIDIA Blackwell Ultra | NVIDIA Hopper |
| Memory | 288GB HBM3e | 80GB HBM3 |
| Common workloads | frontier model training, reasoning inference, memory-intensive AI workloads | large model training, high-throughput inference, distributed fine-tuning |
| Price interpretation | B300 is a distinct 288GB Blackwell Ultra GPU. It is not combined with GB300 systems, and early cloud coverage should be read with the provider count. | H100 SXM is usually the premium H100 shape, with stronger node-level bandwidth than PCIe deployments. |
| Metric | B300 SXM 288GB | H100 SXM5 80GB |
|---|---|---|
| Index median | $7.40/hr | $3.95/hr |
| Community marketplace median | $6.94/hr | $2.69/hr |
| Cheapest current row | $6.94 · RunPod | $1.95 · Shadeform |
| Provider coverage | 4 providers · 18 listings | 5 providers · 44 listings |
| Best use of this page | Use this as a price comparison and sourcing snapshot. Performance, memory size, software support, and availability still need to be checked with the provider. | |
These are published rates, normalized per GPU per hour and labeled by billing mode. The index median uses eligible reliable-tier providers only. Read the methodology →