← Back to community benchmarks

gemma-4-31b-it

M5 Max (40c) · 128 GB · 4bit · 2026-05-22
Performance
32k
tokens
455.3
PP tok/s
14.9
TG tok/s
71976
TTFT (ms)
24.9
Peak mem (GB)
Hardware
Chip M5 Max (40c)
Memory 128 GB
GPU Cores 40
Software
oMLX v0.3.9rc1
macOS macOS 26.5
Context 32,768
Performance by Context Length
Context PP tok/s TG tok/s Peak Mem
1k 661.9 26.4 17.6 GB view
4k 598.8 24.4 20.2 GB view
8k 552.3 21.9 20.7 GB view
16k 506.0 21.2 21.9 GB view
32k 455.3 14.9 24.9 GB current
64k 243.2 8.8 30.9 GB view
128k 194.3 6.8 42.8 GB view
195k 145.9 4.8 57.9 GB view
Batching Results
Batch Size TG tok/s Speedup
26.4 1.00×
20.8 0.79×
25.7 0.97×
29.3 1.11×