← Back to community benchmarks

gemma-4-31b-mxfp8

M4 Pro (16c) · 64 GB · 8bit · 2026-04-03
Performance
8k
tokens
78.3
PP tok/s
6.3
TG tok/s
104634
TTFT (ms)
34.1
Peak mem (GB)
Hardware
Chip M4 Pro (16c)
Memory 64 GB
GPU Cores 16
Software
oMLX v0.3.1
macOS macOS 26.3
Context 8,192
Performance by Context Length
Context PP tok/s TG tok/s Peak Mem
1k 81.2 6.9 32.3 GB view
4k 79.3 6.6 33.9 GB view
8k 78.3 6.3 34.1 GB current
16k 76.1 4.4 35.0 GB view
32k 70.4 1.8 37.3 GB view
64k 52.7 1.0 42.1 GB view
Batching Results
Batch Size TG tok/s Speedup
6.9 1.00×
6.8 0.99×
6.8 0.99×