← Back to community benchmarks

gemma-4-31b-it-bf16

M3 Ultra (80c) · 512 GB · bf16 · 2026-05-06
Performance
16k
tokens
325.4
PP tok/s
9.5
TG tok/s
50344
TTFT (ms)
62.9
Peak mem (GB)
Hardware
Chip M3 Ultra (80c)
Memory 512 GB
GPU Cores 80
Software
oMLX v0.3.8
macOS macOS 26.4.1
Context 16,384
Performance by Context Length
Context PP tok/s TG tok/s Peak Mem
1k 316.7 10.1 58.5 GB view
8k 335.7 9.9 61.7 GB view
16k 325.4 9.5 62.9 GB current
32k 302.3 9.0 65.8 GB view
64k 257.9 8.0 71.8 GB view
Batching Results
Batch Size TG tok/s Speedup
10.1 1.00×
11.9 1.18×
22.9 2.27×
39.3 3.89×