← Back to community benchmarks

Qwen3.5-9B

M3 Ultra (80c) · 512 GB · 4bit · 2026-03-13
Performance
4k
tokens
1,461
PP tok/s
99.4
TG tok/s
2803
TTFT (ms)
7.0
Peak mem (GB)
Hardware
Chip M3 Ultra (80c)
Memory 512 GB
GPU Cores 80
Software
oMLX v0.2.10
macOS macOS 26.3.1
Context 4,096
Performance by Context Length
Context PP tok/s TG tok/s Peak Mem
1k 1,357 104.5 6.7 GB view
4k 1,461 99.4 7.0 GB current
Batching Results
Batch Size TG tok/s Speedup
104.5 1.00×
178.8 1.71×
244.8 2.34×
306.4 2.93×