← Back to community benchmarks

gemma-4-31b-it

M5 Max (40c) · 128 GB · 4bit · 2026-04-03
Performance
4k
tokens
701.6
PP tok/s
19.3
TG tok/s
5838
TTFT (ms)
22.4
Peak mem (GB)
Hardware
Chip M5 Max (40c)
Memory 128 GB
GPU Cores 40
Software
oMLX v0.3.2
macOS macOS 26.4
Context 4,096
Performance by Context Length
Context PP tok/s TG tok/s Peak Mem
1k 698.5 21.3 20.7 GB view
4k 701.6 19.3 22.4 GB current
8k 671.3 18.8 22.6 GB view
16k 582.4 15.4 23.4 GB view
32k 512.7 9.8 25.8 GB view
64k 391.3 6.0 30.5 GB view
128k 281.8 3.3 40.1 GB view
Batching Results
Batch Size TG tok/s Speedup
21.3 1.00×
26.8 1.26×
36.5 1.71×