← Back to community benchmarks

Qwen3.5-4B

M3 Pro (18c) · 36 GB · 4bit · 2026-03-12
Performance
4k
tokens
568.2
PP tok/s
45.3
TG tok/s
7209
TTFT (ms)
4.2
Peak mem (GB)
Hardware
Chip M3 Pro (18c)
Memory 36 GB
GPU Cores 18
Software
oMLX v0.2.8
macOS macOS 26.0.1
Context 4,096
Performance by Context Length
Context PP tok/s TG tok/s Peak Mem
1k 543.3 47.3 4.0 GB view
4k 568.2 45.3 4.2 GB current
Batching Results
Batch Size TG tok/s Speedup
47.3 1.00×
88.8 1.88×
114.7 2.42×