← Back to community benchmarks

Huihui-Qwen3.5-9B-abliterated

M3 Pro (18c) · 36 GB · 4bit · 2026-03-10
Performance
64k
tokens
266.6
PP tok/s
13.3
TG tok/s
245800
TTFT (ms)
9.9
Peak mem (GB)
Hardware
Chip M3 Pro (18c)
Memory 36 GB
GPU Cores 18
Software
oMLX v0.2.7
macOS macOS 26.3.1
Context 65,536
Performance by Context Length
Context PP tok/s TG tok/s Peak Mem
1k 323.4 27.7 5.8 GB view
4k 345.1 27.0 6.0 GB view
8k 342.4 26.1 6.3 GB view
16k 332.5 24.5 6.8 GB view
32k 310.7 21.1 7.8 GB view
64k 266.6 13.3 9.9 GB current
Batching Results
Batch Size TG tok/s Speedup
27.7 1.00×
49.6 1.79×
78.5 2.83×