← Back to community benchmarks

Qwen3.5-4B

M3 (10c) · 16 GB · 4bit · 2026-03-10
Performance
8k
tokens
207.1
PP tok/s
18.4
TG tok/s
39558
TTFT (ms)
4.6
Peak mem (GB)
Hardware
Chip M3 (10c)
Memory 16 GB
GPU Cores 10
Software
oMLX v0.2.6
macOS macOS 26.3.1
Context 8,192
Performance by Context Length
Context PP tok/s TG tok/s Peak Mem
1k 320.6 24.0 3.9 GB view
4k 273.9 21.3 4.2 GB view
8k 207.1 18.4 4.6 GB current
Batching Results
Batch Size TG tok/s Speedup
24.0 1.00×
43.0 1.79×
49.6 2.07×
47.2 1.97×