bestmodel.run / model — gemma-4-12b-it-GGUF (agent view)

category chat · unsloth/gemma-4-12b-it-GGUF · 10 runs in the model record
answer: 50 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
  Radeon AI Pro R9700 32GB           | 4-bit | 58.2 tok/s     | reported | ttft 218ms | vram - GB | ctx 2,048
  RTX 2080 Ti 11GB ×2                | 8-bit | 20.8 tok/s     | reported | ttft 1,394ms | vram 20 GB | ctx 2,048
  RTX 3060 12GB                      | 4-bit | 37 tok/s       | reported | ttft -ms | vram 7.6 GB | ctx 32,768
  RTX 3090 24GB ×2                   | 8-bit | 48.7 tok/s     | measured | ttft 349ms | vram - GB | ctx 131,072
  RTX 5060 Ti 16GB                   | 4-bit | 51.3 tok/s     | reported | ttft 227ms | vram - GB | ctx 2,048
  RTX 5070 12GB                      | 5-bit | 62.8 tok/s     | reported | ttft 168ms | vram - GB | ctx 2,048
  RTX 5070 12GB                      | 6-bit | 51.3 tok/s     | reported | ttft 185ms | vram - GB | ctx 2,048
  Ryzen AI Max 385 24GB              | 4-bit | 8.6 tok/s      | reported | ttft -ms | vram - GB | ctx 2,048