← Back to community benchmarks

gemma-4-31b-it

M3 Ultra (80c) · 512 GB · bf16 · 2026-07-21
Performance
8k
tokens
331.7
PP tok/s
9.8
TG tok/s
24697
TTFT (ms)
63.4
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 324.8 10.1 58.5 GB view
4k 335.6 9.9 60.3 GB view
8k 331.7 9.8 63.4 GB current
16k 317.8 9.6 67.0 GB view
32k 291.2 9.0 70.2 GB view
64k 250.4 8.1 76.1 GB view
Batching Results
Batch Size TG tok/s Speedup
10.1 1.00×
12.0 1.19×
22.9 2.27×
39.5 3.91×