← Back to community benchmarks

Qwen3.5-9B

M3 Pro (18c) · 36 GB · 4bit · 2026-03-21
Performance
8k
tokens
339.0
PP tok/s
24.5
TG tok/s
24166
TTFT (ms)
7.8
Peak mem (GB)
Hardware
Chip M3 Pro (18c)
Memory 36 GB
GPU Cores 18
Software
oMLX v0.2.19
macOS macOS 26.3.1
Context 8,192
Performance by Context Length
Context PP tok/s TG tok/s Peak Mem
1k 345.3 26.5 6.7 GB view
4k 342.9 25.8 7.4 GB view
8k 339.0 24.5 7.8 GB current
Batching Results
Batch Size TG tok/s Speedup
26.5 1.00×
52.6 1.98×
77.7 2.93×
78.1 2.95×