Benchmark result
Qwen3.6-40B-Claude-4.6-Opus-Deckard-Heretic-Uncensored-Thinking-8bit on Apple M5 Max — 8.3 tok/s
Measured with oMLX 0.31.3 on July 14, 2026.
What the model built
Open full screen →The coding scenario asks for a playable game in a single HTML file. This is exactly what the model returned, unedited.
How this run compares
2nd fastest of 2 runs of this model on Apple M5 Max · median 10.1 tok/s.
sort
- 112.0tok/s76.5oMLX
- 28.3tok/s73.8oMLXthis run
Reproduce this run
tooloMLX 0.31.3
modelQwen3.6-40B-Claude-4.6-Opus-Deckard-Heretic-Uncensored-Thinking-8bit
context65,536
clientv0.4.26+97
Set those in oMLX, then:
llm-benchmark benchmark --model "Qwen3.6-40B-Claude-4.6-Opus-Deckard-Heretic-Uncensored-Thinking-8bit" --tool "oMLX"The client prompts for context length and thinking mode, and for the KV cache dtype on Unsloth Studio. Sampling is left at the model default — the values above are what the tool reported using, not overrides the client sent.