Benchmark result
Tiel-Coder-35B-A3B-MLX-oQ6e-MTP on Apple M5 Max — 166.5 tok/s
Measured with oMLX 0.31.4.dev132+g94cdcae13 on October 2, 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; the frame only adds a line that tells this page how tall it is.
How this run compares
1st fastest of 5 runs of this model on Apple M5 Max · median 125.2 tok/s.
sort
- 1166.5tok/s86.3oMLX · oQ6e · KV nonethis run
- 2139.5tok/s80.2oMLX · oQ6e · KV none
- 3125.2tok/s71.7oMLX · oQ6e · KV none
- 4118.0tok/s82.2oMLX · oQ6e · KV tq8
- 5107.7tok/s81.0oMLX · oQ6e · KV none
Tiel-Coder-35B-A3B-MLX-oQ6e-MTP on other hardware
Reproduce this run
tooloMLX 0.31.4.dev132+g94cdcae13
modelTiel-Coder-35B-A3B-MLX-oQ6e-MTP (oQ6e)
context262,144
kv cachenone
thinkingon
think limitnone
samplingtemperature 1 · top_p 0.95 · top_k 20 · repetition_penalty 1
clientv0.4.87+102
Set those in oMLX, then:
llm-benchmark benchmark --model "Tiel-Coder-35B-A3B-MLX-oQ6e-MTP" --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.