Benchmark result
qwen/qwen3.6-35b-a3b on Apple M1 Max — 62.9 tok/s
Measured with LM Studio on November 19, 2026.
Support llm-bench.io
llm-bench.io is free. If it helped you, I'd appreciate your support to cover the infrastructure costs.
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; the frame only adds a line that tells this page how tall it is.
How this run compares
7th fastest of 13 runs of this model on Apple M1 Max · median 62.9 tok/s.
sort
- 186.0tok/s80.6LM Studio · 4bit
- 269.4tok/s76.0LM Studio · 8bit
- 368.7tok/s82.1LM Studio · 8bit
- 467.9tok/s79.9LM Studio · 6bit
- 566.3tok/s78.7LM Studio · 6bit
- 664.3tok/s71.3LM Studio · 4bit
- 762.9tok/s78.6LM Studio · 4bitthis run
- 859.8tok/s79.5LM Studio · Q4_K_M
- 959.7tok/s75.9LM Studio · 4bit
- 1058.2tok/s81.3LM Studio · 4bit
- 1153.9tok/s81.0LM Studio · Q4_K_M
- 1248.2tok/s81.2LM Studio · 6bit
- 1347.5tok/s84.1LM Studio · 6bit
Reproduce this run
toolLM Studio
modelqwen/qwen3.6-35b-a3b (4bit)
context65,536
clientv0.4.99+102
Set those in LM Studio, then:
llm-benchmark benchmark --model "qwen/qwen3.6-35b-a3b" --tool "LM Studio"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.