← Back to community benchmarks

gemma-4-31b-it

M3 Ultra (60c) · 256 GB · 4bit · 2026-04-03
Performance
16k
tokens
241.5
PP tok/s
14.0
TG tok/s
67835
TTFT (ms)
23.4
Peak mem (GB)
Hardware
Chip M3 Ultra (60c)
Memory 256 GB
GPU Cores 60
Software
oMLX v0.3.1
macOS macOS 26.3
Context 16,384
Performance by Context Length
Context PP tok/s TG tok/s Peak Mem
1k 255.9 23.2 20.8 GB view
4k 253.7 20.9 22.4 GB view
8k 250.4 18.8 22.6 GB view
16k 241.5 14.0 23.4 GB current
32k 225.6 10.4 25.8 GB view
64k 196.5 6.6 31.1 GB view
128k 143.9 2.3 40.0 GB view
195k 116.0 1.3 50.1 GB view
Batching Results
Batch Size TG tok/s Speedup
23.2 1.00×
26.4 1.14×
37.0 1.59×
48.0 2.07×