← Back to community benchmarks

Qwen3.5-9B

M5 Max (40c) · 128 GB · 4bit · 2026-07-17
Performance
64k
tokens
762.4
PP tok/s
26.4
TG tok/s
85965
TTFT (ms)
13.7
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 1,059 44.5 5.8 GB view
4k 1,210 43.0 6.5 GB view
8k 1,194 42.2 7.2 GB view
16k 1,143 39.7 8.3 GB view
32k 1,031 35.3 10.0 GB view
64k 762.4 26.4 13.7 GB current
Batching Results
Batch Size TG tok/s Speedup
44.5 1.00×
66.1 1.49×
106.5 2.39×