← Back to community benchmarks

Qwen3.5-9B

M5 Max (40c) · 128 GB · 4bit · 2026-07-17
Performance
64k
tokens
837.9
PP tok/s
29.2
TG tok/s
78217
TTFT (ms)
14.1
Peak mem (GB)
Hardware
Chip M5 Max (40c)
Memory 128 GB
GPU Cores 40
Software
oMLX v0.5.1
macOS macOS 27.0
Context 65,536
Performance by Context Length
Context PP tok/s TG tok/s Peak Mem
1k 2,200 43.3 5.8 GB view
4k 1,202 43.3 6.5 GB view
8k 1,196 41.1 7.0 GB view
16k 1,141 38.6 8.3 GB view
32k 1,021 35.8 10.3 GB view
64k 837.9 29.2 14.1 GB current
Batching Results
Batch Size TG tok/s Speedup
43.3 1.00×
65.6 1.52×
103.1 2.38×