← Back to community benchmarks

gemma-4-31b-it-mxfp8

M3 Ultra (60c) · 256 GB · 8bit · 2026-05-23
Performance
4k
tokens
249.2
PP tok/s
18.0
TG tok/s
16434
TTFT (ms)
33.6
Peak mem (GB)
Hardware
Chip M3 Ultra (60c)
Memory 256 GB
GPU Cores 60
Software
oMLX v0.3.9
macOS macOS 26.5
Context 4,096
Performance by Context Length
Context PP tok/s TG tok/s Peak Mem
1k 245.4 18.4 30.9 GB view
4k 249.2 18.0 33.6 GB current
8k 247.2 17.5 34.1 GB view
16k 238.0 16.5 35.3 GB view
32k 220.8 14.9 38.7 GB view
64k 192.3 12.2 44.9 GB view
128k 150.2 8.1 56.2 GB view
Batching Results
Batch Size TG tok/s Speedup
18.4 1.00×
23.7 1.29×
26.6 1.45×