← Back to community benchmarks

gemma-4-26b-a4b-it-mxfp8

M3 Ultra (80c) · 512 GB · 8bit · 2026-04-15

Performance

4k

tokens

2,036

PP tok/s

76.0

TG tok/s

2012

TTFT (ms)

25.5

Peak mem (GB)

Hardware

Chip M3 Ultra (80c)

Memory 512 GB

GPU Cores 80

Software

oMLX v0.3.5.dev1

macOS macOS 26.4

Context 4,096

Performance by Context Length

Context	PP tok/s	TG tok/s	Peak Mem
1k	1,498	79.7	25.0 GB	view
4k	2,036	76.0	25.5 GB	current

Batching Results

Batch Size	TG tok/s	Speedup
1×	79.7	1.00×
2×	121.9	1.53×
4×	164.5	2.06×