Qwen
Qwen
Qwen/Qwen3.8-27B

Runs

View all benchmark runs for this model family.

	Quant						Actions

	Quant		Actions

	Quant						Actions
M1 Max	q4_k_s	llama.cpp0.3.0-dev	8.1 tok/s	93.7 tok/s	15.50 GB 24%
M3 Max	q4_k_s	llama.cpp0.3.0-dev	12.0 tok/s	172.3 tok/s	16.54 GB 13%
M3 Max	q4_k_s	llama.cpp0.3.0-dev	12.6 tok/s	170.9 tok/s	16.64 GB 13%
M5 Pro	8bit	mlx_lm0.31.3	9.8 tok/s	406.2 tok/s	30.00 GB 47%
M5 Pro	nvfp4	mlx_lm0.31.3	17.7 tok/s	470.8 tok/s	18.00 GB 28%
M4 Max	4bit	mlx_lm0.31.3	19.9 tok/s	197.6 tok/s	18.00 GB 50%
M4 Max	q4_k_s	llama.cppb9960	18.9 tok/s	190.9 tok/s	4.63 GB 13%

	Quant		Actions
M1 Max	q4_k_s	8.1 tok/s
M3 Max	q4_k_s	12.0 tok/s
M3 Max	q4_k_s	12.6 tok/s
M5 Pro	8bit	9.8 tok/s
M5 Pro	nvfp4	17.7 tok/s
M4 Max	4bit	19.9 tok/s
M4 Max	q4_k_s	18.9 tok/s