← Back to community benchmarks

Qwen3.5-9B

M5 Pro (20c) · 64 GB · 4bit · 2026-03-16
Performance
195k
tokens
640.9
PP tok/s
8.7
TG tok/s
312072
TTFT (ms)
19.5
Peak mem (GB)
Hardware
Chip M5 Pro (20c)
Memory 64 GB
GPU Cores 20
Software
oMLX v0.2.12
macOS macOS 26.3.1
Context 200,000
Performance by Context Length
Context PP tok/s TG tok/s Peak Mem
1k 1,563 57.5 6.7 GB view
4k 1,641 55.3 7.0 GB view
8k 1,579 53.8 7.1 GB view
16k 1,512 50.1 7.6 GB view
32k 1,385 44.0 8.7 GB view
64k 1,161 34.9 10.8 GB view
128k 838.7 12.5 15.0 GB view
195k 640.9 8.7 19.5 GB current
Batching Results
Batch Size TG tok/s Speedup
57.5 1.00×
104.8 1.82×
141.3 2.46×
151.5 2.63×