← Back to community benchmarks

gemma-4-31b-it

M3 Max (40c) · 128 GB · 4bit · 2026-05-02
Performance
8k
tokens
153.3
PP tok/s
16.4
TG tok/s
53443
TTFT (ms)
19.6
Peak mem (GB)
Hardware
Chip M3 Max (40c)
Memory 128 GB
GPU Cores 40
Software
oMLX v0.3.8
macOS macOS 26.4
Context 8,192
Performance by Context Length
Context PP tok/s TG tok/s Peak Mem
1k 170.0 18.4 17.6 GB view
4k 163.0 16.5 20.2 GB view
8k 153.3 16.4 19.6 GB current
16k 131.9 9.2 21.9 GB view
32k 116.5 9.8 24.9 GB view
64k 107.6 8.8 30.9 GB view
128k 86.2 3.7 42.8 GB view
195k 75.0 2.0 47.1 GB view
Batching Results
Batch Size TG tok/s Speedup
18.4 1.00×
26.8 1.46×
33.2 1.80×
34.6 1.88×