Google
Google
Google/Gemma 4 E4B IT

Runs

View all benchmark runs for this model family.

	Quant
M4 Pro	4bit	mlx_lm0.31.2	64.4 tok/s	1,230.0 tok/s	5.78 GB 12%
M4 Max	4bit	mlx_lm0.31.2	98.7 tok/s	2,073.2 tok/s	6.06 GB 5%
M4 Max	q6_k	llama.cppb8680	62.1 tok/s	1,117.5 tok/s	7.70 GB 21%
M2 Pro	8bit	mlx_lm0.31.2	28.7 tok/s	480.8 tok/s	9.17 GB 57%
M2 Pro	4bit	mlx_lm0.31.2	44.7 tok/s	480.7 tok/s	5.72 GB 36%
M2 Pro	q4_k_m	llama.cppb8680	34.1 tok/s	438.3 tok/s	2.75 GB 17%
M1	4bit	mlx_lm0.31.2	17.2 tok/s	225.4 tok/s	6.09 GB 38%
M2 Pro	q4_k_m	llama.cppb8680	34.9 tok/s	442.3 tok/s	1.66 GB 10%
GeForce RTX 5070 Ti	q4_k_m	llama.cppb1	136.4 tok/s	8,064.1 tok/s	5.00 GB 22%
M1 Max	q4_k_m	llama.cppb8680	41.7 tok/s	635.2 tok/s	4.05 GB 13%