← Back to community benchmarks

gemma-4-31B-it-oQ4

M3 Ultra (80c) · 512 GB · 4bit · 2026-05-13

Performance

1k

tokens

315.4

PP tok/s

29.2

TG tok/s

3247

TTFT (ms)

18.1

Peak mem (GB)

Hardware

Chip M3 Ultra (80c)

Memory 512 GB

GPU Cores 80

Software

oMLX v0.3.9.dev2

macOS macOS 26.4

Context 1,024

Performance by Context Length

Context	PP tok/s	TG tok/s	Peak Mem
1k	315.4	29.2	18.1 GB	current
4k	332.2	28.4	20.7 GB	view
8k	330.9	27.1	21.3 GB	view
16k	308.3	25.2	22.4 GB	view
32k	284.3	22.2	25.4 GB	view
64k	235.7	17.4	31.3 GB	view