NVIDIA H100 NVL vs M3 Ultra 96GB for local LLMs
1 report on the NVIDIA H100 NVL and 11 on the M3 Ultra 96GB
Which is faster for local LLMs?
On gpt-oss 117B · 5.1B active at 4-bit, the NVIDIA H100 NVL has 177.7 generation tokens/s (one run); the M3 Ultra 96GB has 73.0 generation tokens/s (one run); prompt processing 347.0 tokens/s (one run). Those are community reports whose engines and settings differ, so the gap is a rough sign, not a controlled measure of the two cards. Memory bandwidth, which caps how fast a card can write, is 4.8x higher on the NVIDIA H100 NVL (3,900 GB/s on the NVIDIA H100 NVL, 819 on the M3 Ultra 96GB).
- Memory
- 94GB
- Memory bandwidth
- 3,900 GB/s
- FP16 compute
- 835.5 TFLOPS
- Reports
- 1
- Memory
- 96GB unified
- Memory bandwidth
- 819 GB/s
- FP16 compute
- 21.3 TFLOPS
- Reports
- 11
Measured on both
Generation and prompt processing speeds from plain runs (one device, one request, no speculative decoding, the whole model in the device's memory) of the same model size at the same bit class. Each side shows the engines and reported context behind its median. Context can mean a configured window or prompt depth, so these are community comparisons with differing settings. How to read these
| Model | NVIDIA H100 NVL | M3 Ultra 96GB |
|---|---|---|
| gpt-oss 117B · 5.1B active4-bit | 177.7one runllama.cpp32K context | 73.0one runPP 347.0 (one run)MLXcontext not stated |
Estimated by the calculator
The quant the calculator recommends for each card and its estimated speed at a 32,768-token context with 32 GB of system RAM, from memory bandwidth and the model's shape. Where a model was also measured above, the measurement wins.
| Model | NVIDIA H100 NVL | M3 Ultra 96GB |
|---|---|---|
| gpt-oss 117B · 5.1B active | 222.8 t/sQ5_K_S | 68.5 t/sQ3_K_M |
| Qwen3.8 27B | 39.7 t/sFP16 | 7.9 t/sFP16 |
| Qwen3.8 125B · 6B active | 217.3 t/sQ4_K_M | 55.1 t/sQ3_K_M |
| Qwen3.6 35B · 3B active | 123.0 t/sFP16 | 57.1 t/sQ8_0 |
| DeepSeek V4 Flash 284B · 13B active | 9.6 t/sQ2_K, experts in RAM | doesn't fit at these settings |
| Qwen3.6 27B | 39.7 t/sFP16 | 7.9 t/sFP16 |
| Gemma 4 26B · 4B active | 107.3 t/sFP16 | 19.0 t/sFP16 |
Change the context, RAM or model in the calculator: NVIDIA H100 NVL · M3 Ultra 96GB
Every report on each card, with its source: NVIDIA H100 NVL · M3 Ultra 96GB