← Back to community benchmarks

Qwen3.5-9B-OptiQ

M1 Ultra (48c) · 64 GB · 4bit · 2026-03-13
Performance
4k
tokens
454.3
PP tok/s
66.4
TG tok/s
9016
TTFT (ms)
7.0
Peak mem (GB)
Hardware
Chip M1 Ultra (48c)
Memory 64 GB
GPU Cores 48
Software
oMLX v0.2.9
macOS macOS 26.3.1
Context 4,096
Performance by Context Length
Context PP tok/s TG tok/s Peak Mem
1k 410.2 69.2 6.8 GB view
4k 454.3 66.4 7.0 GB current
8k 458.0 60.6 7.2 GB view
16k 453.8 52.7 7.7 GB view
32k 435.5 41.2 8.8 GB view
Batching Results
Batch Size TG tok/s Speedup
69.2 1.00×
86.3 1.25×
101.6 1.47×
107.6 1.55×