← Back to community benchmarks

gemma-4-E4B-it-PARO

M5 Max (40c) · 128 GB · 4bit · 2026-05-25

Performance

128k

tokens

3,600

PP tok/s

40.5

TG tok/s

36404

TTFT (ms)

13.9

Peak mem (GB)

Hardware

Chip M5 Max (40c)

Memory 128 GB

GPU Cores 40

Software

oMLX v0.3.10

macOS macOS 26.5

Context 131,072

Performance by Context Length

Context	PP tok/s	TG tok/s	Peak Mem
1k	2,877	112.2	10.6 GB	view
64k	4,847	59.1	12.1 GB	view
128k	3,600	40.5	13.9 GB	current

Batching Results

Batch Size	TG tok/s	Speedup
1×	112.2	1.00×
2×	194.8	1.74×
4×	305.8	2.73×