gemma4 12B Q4_KM
3 records across 2 hardware.
| hardware | model | quant | backend | tok/s | pp tok/s | W | ttft s | ctx/tg | date | class | flags | source |
|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Arc B580 | gemma4 12B Q4_KM | as-published | llama.cpp (SYCL) | 41.34 | 1320.02 | – | – | tg128 | 2026-06-13 | community | – | source |
| Arc B580 | gemma4 12B Q4_KM | as-published | llama.cpp (SYCL) | 39.96 | 1062.58 | – | – | tg128 | 2026-06-13 | community | – | source |
| RTX 3060 12GB | gemma4 12B Q4_KM | Q4_K_M | llama.cpp (CUDA) | 33.34 | 609.26 | – | – | – | 2026-06-09 | community | – | source |