← Back to community benchmarks

Qwen3.5-9B

M3 Ultra (80c) · 512 GB · 8bit · 2026-03-15
Performance
4k
tokens
1,457
PP tok/s
62.1
TG tok/s
2811
TTFT (ms)
11.0
Peak mem (GB)
Hardware
Chip M3 Ultra (80c)
Memory 512 GB
GPU Cores 80
Software
oMLX v0.2.10
macOS macOS 15.3
Context 4,096
Performance by Context Length
Context PP tok/s TG tok/s Peak Mem
1k 1,424 64.1 10.7 GB view
4k 1,457 62.1 11.0 GB current
8k 1,443 59.7 11.2 GB view
16k 1,395 54.7 11.7 GB view
Batching Results
Batch Size TG tok/s Speedup
64.1 1.00×
112.0 1.75×
166.8 2.60×
210.1 3.28×