← Back to community benchmarks

gemma-4-E4B-it-heretic-oQ6

M3 (10c) · 16 GB · 6bit · 2026-05-18
Performance
32k
tokens
247.6
PP tok/s
16.4
TG tok/s
132329
TTFT (ms)
8.0
Peak mem (GB)
Hardware
Chip M3 (10c)
Memory 16 GB
GPU Cores 10
Software
oMLX v0.3.9.dev2
macOS macOS 26.5
Context 32,768
Performance by Context Length
Context PP tok/s TG tok/s Peak Mem
1k 524.3 49.3 6.9 GB view
4k 528.0 44.8 7.0 GB view
8k 483.6 43.6 7.2 GB view
16k 250.3 12.5 7.4 GB view
32k 247.6 16.4 8.0 GB current
64k 182.1 11.7 9.2 GB view
Batching Results
Batch Size TG tok/s Speedup
49.3 1.00×
12.7 0.26×
20.1 0.41×
21.1 0.43×