M3 Pro (18 GPU)
3 records across 1 models (quants: F16, Q4_0, Q8_0).
| hardware | model | quant | backend | tok/s | pp tok/s | W | ttft s | ctx/tg | date | class | flags | source |
|---|---|---|---|---|---|---|---|---|---|---|---|---|
| M3 Pro (18 GPU) | llama 7B | Q4_0 | llama.cpp (Metal) | 30.74 | 341.67 | – | – | – | 2023-11-22 | community | – | source |
| M3 Pro (18 GPU) | llama 7B | Q8_0 | llama.cpp (Metal) | 17.53 | 344.66 | – | – | – | 2023-11-22 | community | – | source |
| M3 Pro (18 GPU) | llama 7B | F16 | llama.cpp (Metal) | 9.89 | 357.45 | – | – | – | 2023-11-22 | community | – | source |
All rows cite their source. Provenance: community 40018 sourced 1085