Google
Google
Google/Gemma 4 E2B

Runs

View all benchmark runs for this model family.

	Quant						Actions

	Quant		Actions

	Quant						Actions
M3 Pro	6bit	mlx_lm0.31.2	54.8 tok/s	2,537.6 tok/s	5.15 GB 29%
M3 Pro	6bit	mlx_lm0.31.2	54.7 tok/s	2,530.9 tok/s	5.37 GB 30%
M2 Max	5bit	mlx_lm0.31.2	91.9 tok/s	3,309.2 tok/s	4.75 GB 7%
M1 Max	nvfp4	mlx_lm0.31.2	72.9 tok/s	2,279.7 tok/s	4.72 GB 7%
M1 Max	mxfp4	mlx_lm0.31.2	70.6 tok/s	1,965.1 tok/s	4.62 GB 7%
M1 Max	8bit	mlx_lm0.31.2	68.7 tok/s	1,939.8 tok/s	6.16 GB 10%
M1 Ultra	4bit	mlx_lm0.31.2	102.9 tok/s	4,655.2 tok/s	4.24 GB 7%
M4 Max	4bit	mlx_lm0.31.2	167.6 tok/s	8,158.4 tok/s	4.21 GB 3%

	Quant		Actions
M3 Pro	6bit	54.8 tok/s
M3 Pro	6bit	54.7 tok/s
M2 Max	5bit	91.9 tok/s
M1 Max	nvfp4	72.9 tok/s
M1 Max	mxfp4	70.6 tok/s
M1 Max	8bit	68.7 tok/s
M1 Ultra	4bit	102.9 tok/s
M4 Max	4bit	167.6 tok/s