bestmodel.run / model — Qwen3.6-35B-A3B-GGUF (agent view)

category chat · unsloth/Qwen3.6-35B-A3B-GGUF · 271 runs in the model record
answer: 105.4 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
  B200 192GB                         | 4-bit | 199.9 tok/s    | reported | ttft 187ms | vram 21.7 GB | ctx 32,768
  BC-250 8GB                         | 2-bit | 73.1 tok/s     | measured | ttft 9,773ms | vram - GB | ctx 12,554
  CMP 170HX 64GB                     | 8-bit | 110.3 tok/s    | measured | ttft 12,654ms | vram - GB | ctx 131,338
  M4 Pro 48GB                        | 4-bit | 47.1 tok/s     | reported | ttft -ms | vram - GB | ctx 640
  M5 32GB                            | 4-bit | 34.3 tok/s     | reported | ttft -ms | vram - GB | ctx 64,000
  Radeon 890M 32GB                   | 4-bit | 20.1 tok/s     | reported | ttft 1,530ms | vram - GB | ctx 2,048
  Radeon AI Pro R9700 32GB           | 4-bit | 78.6 tok/s     | reported | ttft 178ms | vram - GB | ctx 2,048
  Radeon RX 7900 XTX 24GB            | 4-bit | 57.6 tok/s     | reported | ttft -ms | vram 19.5 GB | ctx 640
  Radeon RX 9070 16GB                | 3-bit | 76 tok/s       | reported | ttft 360ms | vram 15.5 GB | ctx 65,536
  RTX 3060 12GB                      | 2-bit | 63.4 tok/s     | measured | ttft 613ms | vram - GB | ctx 170,240
  RTX 3060 12GB ×2                   | 4-bit | 78.2 tok/s     | measured | ttft 224ms | vram 21.6 GB | ctx 98,304
  RTX 3070 8GB                       | 4-bit | 41.5 tok/s     | reported | ttft 4,700ms | vram 6.7 GB | ctx 131,072
  RTX 3090 24GB ×2                   | 3-bit | 125.9 tok/s    | reported | ttft -ms | vram 10.1 GB | ctx 131,072
  RTX 3090 24GB ×2                   | 4-bit | 124 tok/s      | reported | ttft -ms | vram 22.5 GB | ctx 1,048,576
  RTX 3090 24GB ×2                   | 6-bit | 40.3 tok/s     | reported | ttft 463ms | vram 19.8 GB | ctx 32,768
  RTX 4070 12GB                      | 4-bit | 28.4 tok/s     | reported | ttft 772ms | vram - GB | ctx 2,048
  RTX 4090 24GB                      | 4-bit | 165.1 tok/s    | reported | ttft -ms | vram 23,691 GB | ctx 64,000
  RTX 5060 Ti 16GB ×2                | 4-bit | 94.4 tok/s     | reported | ttft 81ms | vram - GB | ctx 8,192
  RTX 5090 32GB                      | 4-bit | 324.4 tok/s    | reported | ttft 112ms | vram 22.6 GB | ctx 32,768
  Ryzen AI Max 385 24GB              | 4-bit | 19.2 tok/s     | reported | ttft 1,530ms | vram - GB | ctx 2,048
  Ryzen AI Max 395 64GB              | 4-bit | 60.5 tok/s     | reported | ttft 963ms | vram 21.5 GB | ctx 8,192
  Tesla P100 32GB ×2                 | 16-bit | 67 tok/s       | measured | ttft 3,541ms | vram - GB | ctx 2,048
  Tesla P100-PCIE-16GB 16GB          | 3-bit | 24.2 tok/s     | reported | ttft -ms | vram - GB | ctx 2,048
  Tesla P100-PCIE-16GB 16GB ×2       | 5-bit | 63.9 tok/s     | reported | ttft -ms | vram - GB | ctx 2,048
  Tesla V100 32GB                    | 4-bit | 58.3 tok/s     | measured | ttft 71,447ms | vram 21.5 GB | ctx 262,144