← Back to community benchmarks

gemma-4-E4B-it-PARO

M5 Max (40c) · 128 GB · 4bit · 2026-05-29
Performance
8k
tokens
4,947
PP tok/s
90.6
TG tok/s
1656
TTFT (ms)
10.8
Peak mem (GB)
Hardware
Chip M5 Max (40c)
Memory 128 GB
GPU Cores 40
Software
oMLX v0.3.12
macOS macOS 26.6
Context 8,192
Performance by Context Length
Context PP tok/s TG tok/s Peak Mem
1k 2,789 100.2 10.6 GB view
4k 4,508 95.1 10.6 GB view
8k 4,947 90.6 10.8 GB current
64k 4,635 52.7 12.6 GB view
128k 3,331 34.9 14.9 GB view
Batching Results
Batch Size TG tok/s Speedup
100.2 1.00×
135.5 1.35×
226.3 2.26×