Google
Google
Google/Gemma 4 26B A4B IT

Quantizations

Quant	Quantized by	Size	Decode	Prefill	Score
ud-iq3_xxs	Unsloth Unsloth	10.4 GB	N/A	N/A	N/A
mxfp4	MLX Community MLX Community	13.8 GB	N/A	N/A	N/A
nvfp4	MLX Community MLX Community	14.5 GB	N/A	N/A	N/A
4bit	MLX Community MLX Community	14.5 GB	N/A	N/A	N/A
ud-4bit	Unsloth Unsloth	15.0 GB	N/A	N/A	N/A
q4_k_m	LM Studio LM Studio	15.6 GB	N/A	N/A	N/A
q4_k_m	ggml ggml	15.6 GB	57.0 tok/s	700.5 tok/s	Runs well
ud-q4_k_m	Unsloth Unsloth	15.7 GB	N/A	N/A	N/A
ud-q4_k_xl	Unsloth Unsloth	16.0 GB	48.5 tok/s	705.5 tok/s	Runs ok
5bit	MLX Community MLX Community	17.4 GB	N/A	N/A	N/A
q6_k	Bartowski Bartowski	21.3 GB	46.6 tok/s	682.1 tok/s	Runs well
q8_0	MLX Community MLX Community	25.0 GB	43.7 tok/s	699.4 tok/s	Runs ok
8bit	MLX Community MLX Community	26.0 GB	N/A	N/A	N/A

Results include trials with 4,096 input tokens and 1,024 output tokens only.

28 devices