← Back to community benchmarks

Qwen3.5-9B

M4 (10c) · 24 GB · 4bit · 2026-03-11
Performance
16k
tokens
198.6
PP tok/s
17.4
TG tok/s
82495
TTFT (ms)
7.5
Peak mem (GB)
Hardware
Chip M4 (10c)
Memory 24 GB
GPU Cores 10
Software
oMLX v0.2.7
macOS macOS 15.6
Context 16,384
Performance by Context Length
Context PP tok/s TG tok/s Peak Mem
1k 220.1 20.7 6.5 GB view
4k 223.4 19.9 6.8 GB view
8k 212.7 18.6 7.0 GB view
16k 198.6 17.4 7.5 GB current
32k 185.4 14.4 8.5 GB view
64k 160.9 11.3 10.6 GB view
Batching Results
Batch Size TG tok/s Speedup
20.7 1.00×
40.4 1.95×
50.4 2.43×
51.1 2.47×