4-bit
5 records across 3 hardware.
| hardware | model | quant | backend | tok/s | pp tok/s | W | ttft s | ctx/tg | date | class | flags | source |
|---|---|---|---|---|---|---|---|---|---|---|---|---|
| M4 Max | 4-bit | 4bit | oMLX | 28.0 | 247.2 | – | – | 1024 | 2026-10-14 | community | – | source |
| M4 Max | 4-bit | 4bit | oMLX | 25.8 | 240.2 | – | – | 4096 | 2026-10-14 | community | – | source |
| M2 Max | 4-bit | 4bit | oMLX | 21.0 | 144.7 | – | – | 1024 | 2026-10-10 | community | – | source |
| M2 Max | 4-bit | 4bit | oMLX | 20.4 | 132.8 | – | – | 4096 | 2026-10-10 | community | – | source |
| M6 | 4-bit | 4bit | oMLX | 9.7 | 308.0 | – | – | 4096 | 2026-10-18 | community | – | source |