← Back to community benchmarks

gemma-3-4b-it-qat

M3 Ultra (80c) · 512 GB · 4bit · 2026-05-16

Performance

1k

tokens

1,614

PP tok/s

149.5

TG tok/s

635

TTFT (ms)

3.2

Peak mem (GB)

Hardware

Chip M3 Ultra (80c)

Memory 512 GB

GPU Cores 80

Software

oMLX v0.3.8

macOS macOS 26.4.1

Context 1,024

Performance by Context Length

Context	PP tok/s	TG tok/s	Peak Mem
1k	1,614	149.5	3.2 GB	current
4k	2,427	146.7	3.6 GB	view

Batching Results

Batch Size	TG tok/s	Speedup
1×	149.5	1.00×
2×	254.6	1.70×
4×	362.3	2.42×