← Back to community benchmarks

gemma-4-31b-it

M3 Ultra (80c) · 512 GB · 8bit · 2026-07-21
Performance
8k
tokens
327.5
PP tok/s
17.1
TG tok/s
25013
TTFT (ms)
35.0
Peak mem (GB)
Hardware
Chip M3 Ultra (80c)
Memory 512 GB
GPU Cores 80
Software
oMLX v0.5.1
macOS macOS 26.3
Context 8,192
Performance by Context Length
Context PP tok/s TG tok/s Peak Mem
1k 348.1 17.9 31.8 GB view
4k 339.4 17.5 33.6 GB view
8k 327.5 17.1 35.0 GB current
16k 313.2 16.2 37.8 GB view
32k 283.8 14.8 43.4 GB view
64k 240.4 12.6 52.0 GB view
Batching Results
Batch Size TG tok/s Speedup
17.9 1.00×
34.1 1.91×
62.5 3.49×
86.3 4.82×