bestmodel.run / model — gemma-4-31B-it (agent view) category chat · google/gemma-4-31B-it · 32 runs in the model record answer: 24.8 tok/s (model median) Honesty ladder: measured > reported > extrapolated > formula > no data yet. Empty fields stay empty — never filled with an estimate. rig | quant | median result | basis | ttft | peak vram | max context Arc Pro B70 32GB | 4-bit | 16.4 tok/s | measured | ttft -ms | vram - GB | ctx 131,072 M1 Max 64GB | 8-bit | 8.4 tok/s | reported | ttft 287ms | vram - GB | ctx 2,048 Radeon AI Pro R9700 32GB | 4-bit | 28.8 tok/s | reported | ttft -ms | vram - GB | ctx 2,048 Radeon AI Pro R9700 32GB ×3 | 8-bit | 14.2 tok/s | measured | ttft 1,089ms | vram 45.7 GB | ctx 794 Radeon AI Pro R9700 57GB | 4-bit | 3.3 tok/s | reported | ttft 8,092ms | vram - GB | ctx 2,048 RTX 3090 24GB ×2 | 4-bit | 135.1 tok/s | reported | ttft 71ms | vram 45.1 GB | ctx 2,048 RTX 3090 24GB ×2 | 8-bit | 87.7 tok/s | reported | ttft 110ms | vram 21.6 GB | ctx 8,192 RTX 5090 32GB | 4-bit | 68.5 tok/s | reported | ttft 115ms | vram 17.8 GB | ctx 8,192 RTX 5090 32GB | 6-bit | 46.5 tok/s | reported | ttft 158ms | vram 27.3 GB | ctx 8,192 RTX 5090 32GB | 8-bit | 38 tok/s | reported | ttft 250ms | vram 64 GB | ctx 2,048 RTX PRO 6000 Blackwell 96GB | 16-bit | 40 tok/s | reported | ttft 55ms | vram - GB | ctx 2,048 RTX PRO 6000 Blackwell 96GB ×4 | 16-bit | 98.2 tok/s | reported | ttft 43ms | vram - GB | ctx 2,048 RTX PRO 6000 Blackwell Workstation Edition 96GB | 8-bit | 58.6 tok/s | reported | ttft 68ms | vram - GB | ctx 131,072 RTX PRO 6000 Blackwell Workstation Edition 96GB | 16-bit | 47.4 tok/s | measured | ttft 26,513ms | vram 86.1 GB | ctx 131,072 Ryzen AI Max 395 128GB | 4-bit | 10.7 tok/s | reported | ttft -ms | vram 25.4 GB | ctx 16,384 Tesla P100-PCIE-16GB 16GB ×2 | 5-bit | 15.3 tok/s | reported | ttft -ms | vram - GB | ctx 2,048