H200 vs RTX 3090
For running local LLMs · 79 reports across 12 models
Highest reported generation speed by model family
These maxima come from different community setups. Model size, quantization, GPU count, offloading and concurrency may differ. They do not establish which GPU is faster for the same workload. Check the comparison conditions →
| Model | H200 | RTX 3090 |
|---|---|---|
| Qwen3.8up to 2400B | — | 382.0n=45 |
| Qwen3.6up to 35B | — | 104.0n=12 |
| DeepSeek V4 Flashup to 285B | 4.8n=2 | 60.0n=9 |
| GLM-5.2up to 744B | — | 7.3n=2 |
| Qwen2.5up to 32B | — | 70.0n=2 |
| DeepSeek V3 | — | —n=1 |
| DeepSeek V4up to 284B | — | 15.0n=1 |
| GLM-4.5-Airup to 106B | — | —n=1 |
| GLM-5.3 | — | 55.0n=1 |
| HobbyLMup to 1B | —n=1 | — |
| Museup to 30B | — | 94.0n=1 |
| Qwen3up to 235B | — | 7.5n=1 |