granite-4.2-8b
8 records across 3 hardware.
| hardware | model | quant | backend | tok/s | pp tok/s | W | ttft s | ctx/tg | date | class | flags | source |
|---|---|---|---|---|---|---|---|---|---|---|---|---|
| M4 Max | granite-4.2-8b | q8 | oMLX | 49.4 | 796.7 | – | – | 1024 | 2026-10-13 | community | – | source |
| M4 Max | granite-4.2-8b | q8 | oMLX | 46.9 | 771.7 | – | – | 4096 | 2026-10-13 | community | – | source |
| RTX 5060 Ti | granite-4.2-8b 8B | Q8_0 | llama.cpp (Vulkan) | 42.1 | – | – | – | tg128 | 2026-10-05 | community | – | source |
| RTX 5060 Ti | granite-4.2-8b 8B | Q8_0 | llama.cpp (Vulkan) | 41.9 | – | – | – | tg128 | 2026-10-05 | community | – | source |
| M6 | granite-4.2-8b | 4bit | oMLX | 28.7 | 998.7 | – | – | 1024 | 2026-10-14 | community | – | source |
| M6 | granite-4.2-8b | 4bit | oMLX | 28.6 | 949.7 | – | – | 1024 | 2026-10-14 | community | – | source |
| RTX 5060 Ti | granite-4.2-8b 8B | Q8_0 | llama.cpp (Vulkan) | – | 2955 | – | – | – | 2026-10-05 | community | – | source |
| RTX 5060 Ti | granite-4.2-8b 8B | Q8_0 | llama.cpp (Vulkan) | – | 2437 | – | – | – | 2026-10-05 | community | – | source |