← Back to community benchmarks

gemma-4-26B-A4B-it-heretic-msq-2.6bit

M4 (10c) · 24 GB · 4bit · 2026-05-12

Performance

1k

tokens

360.8

PP tok/s

30.6

TG tok/s

2838

TTFT (ms)

9.1

Peak mem (GB)

Hardware

Chip M4 (10c)

Memory 24 GB

GPU Cores 10

Software

oMLX v0.3.6

macOS macOS 15.7.2

Context 1,024

Batching Results

Batch Size	TG tok/s	Speedup
1×	30.6	1.00×
2×	44.0	1.44×
4×	50.3	1.64×