← Back to community benchmarks

Gemma-4-12B-it-MXFP8-CRACK

M4 Pro (20c) · 64 GB · unknown · 2026-06-05
Performance
8k
tokens
232.7
PP tok/s
14.1
TG tok/s
35207
TTFT (ms)
14.6
Peak mem (GB)
Hardware
Chip M4 Pro (20c)
Memory 64 GB
GPU Cores 20
Software
oMLX v0.4.2.dev2
macOS macOS 26.6
Context 8,192
Performance by Context Length
Context PP tok/s TG tok/s Peak Mem
1k 257.0 14.4 13.4 GB view
4k 249.5 13.8 14.5 GB view
8k 232.7 14.1 14.6 GB current
16k 226.5 13.7 15.2 GB view
32k 197.6 13.2 15.8 GB view
64k 169.1 10.3 17.6 GB view
128k 138.4 11.6 21.2 GB view
Batching Results
Batch Size TG tok/s Speedup
14.4 1.00×
13.5 0.94×
24.0 1.67×
27.6 1.92×