Qwen
Qwen
Qwen/Qwen3.5-27B

Runs

View all benchmark runs for this model family.

	Quant
M4 Max	q4_k_m	llama.cppb8680	17.3 tok/s	189.3 tok/s	11.94 GB 33%
M4 Max	q6_k	llama.cppb8680	14.2 tok/s	191.2 tok/s	6.51 GB 18%
M4 Pro	q4_k_m	llama.cppb8793	12.1 tok/s	118.1 tok/s	16.88 GB 26%
M3 Max	8bit	mlx_lm0.31.1	11.3 tok/s	174.3 tok/s	31.00 GB 24%
M4 Pro	4bit	mlx_lm0.31.1	15.1 tok/s	126.3 tok/s	18.00 GB 38%
M2	4bit	mlx_lm0.31.1	3.6 tok/s	21.0 tok/s	18.00 GB 75%
M4 Max	q6_k	llama.cppb8500	10.0 tok/s	134.9 tok/s	23.60 GB 37%
M5 Max	4bit	mlx_lm0.31.1	20.0 tok/s	472.0 tok/s	18.00 GB 14%
M1 Max	4bit	mlx_lm0.31.0	7.5 tok/s	38.7 tok/s	18.00 GB 28%