H100 80GB vs M3 Ultra 512GB
For running local LLMs · 10 reports across 6 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 | H100 80GB | M3 Ultra 512GB |
|---|---|---|
| DeepSeek V4 Flashup to 285B | 193.0n=1 | 20.0n=3 |
| DeepSeek V4.1 Flashup to 552B | — | 31.3n=2 |
| Gemma 4up to 31B | 125.3n=1 | — |
| Qwen3.5up to 122B | —n=1 | — |
| Qwen3.6up to 35B | 45.0n=1 | — |
| Qwen3.8up to 2400B | —n=1 | — |