← Back to community benchmarks

gemma4H-26B-oQ8e

M3 Ultra (80c) · 512 GB · 8bit · 2026-07-17

Performance

32k

tokens

1,848

PP tok/s

56.8

TG tok/s

17728

TTFT (ms)

30.2

Peak mem (GB)

Hardware

Chip M3 Ultra (80c)

Memory 512 GB

GPU Cores 80

Software

oMLX v0.5.1

macOS macOS 26.5.2

Context 32,768

Performance by Context Length

Context	PP tok/s	TG tok/s	Peak Mem
1k	2,422	76.7	25.8 GB	view
4k	2,339	74.3	26.2 GB	view
8k	2,201	70.6	27.0 GB	view
16k	2,066	65.9	28.1 GB	view
32k	1,848	56.8	30.2 GB	current
64k	1,506	43.1	30.4 GB	view
128k	1,066	28.6	34.4 GB	view