← Back to community benchmarks

gemma-4-31b-it

M4 Max (40c) · 128 GB · unknown · 2026-05-16
Performance
16k
tokens
213.1
PP tok/s
7.3
TG tok/s
76892
TTFT (ms)
62.9
Peak mem (GB)
Hardware
Chip M4 Max (40c)
Memory 128 GB
GPU Cores 40
Software
oMLX v0.3.8
macOS macOS 26.3
Context 16,384
Performance by Context Length
Context PP tok/s TG tok/s Peak Mem
1k 215.6 7.7 58.5 GB view
4k 223.4 7.6 60.8 GB view
8k 221.0 7.5 62.1 GB view
16k 213.1 7.3 62.9 GB current
32k 197.9 7.1 65.8 GB view
64k 172.5 6.6 71.8 GB view
128k 134.7 5.7 78.1 GB view
195k 108.8 5.1 88.1 GB view
Batching Results
Batch Size TG tok/s Speedup
7.7 1.00×
7.5 0.97×
14.7 1.91×
28.1 3.65×