Side-by-side GPU comparison: specs, memory, compute performance, and live cloud pricing.
Verdict
B200 has 96 GB more VRAM, making it better suited for large models and long context windows. For compute-bound workloads like training, B200 delivers 2.3× higher FP16 throughput. At $2.29/hr vs $5.49/hr, GH200 is the more cost-efficient choice for inference. B200 supports a broader range of models (659 vs 637 from this catalog), giving more flexibility.
Specifications
| B200 | GH200 | |
|---|---|---|
| VRAM | 192 GB | 96 GB |
| VRAM Type | HBM3e | HBM3 |
| Memory Bandwidth | 8.0 TB/s | 4.0 TB/s |
| FP16 Performance | 2250 TFLOPS | 990 TFLOPS |
| Manufacturer | NVIDIA | NVIDIA |
| FP8 Support | Yes | Yes |
| FP4 Support | Yes | No |
Price / Performance
Based on cheapest single-GPU on-demand pricing. Lower $/TFLOP = better compute value; lower $/GB = better memory value.
| B200 | GH200 | |
|---|---|---|
| $/hr (cheapest) | $5.49 | $2.29✓ best |
| $/TFLOP (compute value) | $0.0024 | $0.0023✓ best |
| $/GB VRAM (memory value) | $0.0286 | $0.0239✓ best |
Cloud Pricing
Cheapest on-demand price per provider (single GPU).
Model Compatibility
Models from the catalog that fit on each GPU, grouped by required precision.
B200 (659 models)
GH200 (637 models)
You might also compare…
Pricing data refreshed hourly · Last updated April 11, 2026 · Browse all comparisons