← Back to community benchmarks

gemma-4-31b-it

M2 Ultra (60c) · 128 GB · 4bit · 2026-07-19
Performance
1k
tokens
154.6
PP tok/s
23.2
TG tok/s
6624
TTFT (ms)
17.6
Peak mem (GB)
Hardware
Chip M2 Ultra (60c)
Memory 128 GB
GPU Cores 60
Software
oMLX v0.5.1
macOS macOS 26.5
Context 1,024
Performance by Context Length
Context PP tok/s TG tok/s Peak Mem
1k 154.6 23.2 17.6 GB current
4k 150.8 22.4 19.3 GB view
8k 147.3 21.3 21.6 GB view
16k 143.1 19.6 24.5 GB view
32k 136.0 16.8 27.4 GB view
64k 120.8 13.0 30.9 GB view
128k 93.4 8.9 43.7 GB view
195k 75.6 6.7 61.3 GB view
Batching Results
Batch Size TG tok/s Speedup
23.2 1.00×
37.6 1.62×
51.7 2.23×
57.2 2.47×