← Back to community benchmarks

Qwen3.5-4B

M2 (10c) · 8 GB · 4bit · 2026-03-16
Performance
8k
tokens
193.8
PP tok/s
18.2
TG tok/s
42262
TTFT (ms)
4.6
Peak mem (GB)
Hardware
Chip M2 (10c)
Memory 8 GB
GPU Cores 10
Software
oMLX v0.2.10
macOS macOS 15.7.5
Context 8,192
Performance by Context Length
Context PP tok/s TG tok/s Peak Mem
1k 199.3 32.1 3.9 GB view
4k 198.0 29.7 4.2 GB view
8k 193.8 18.2 4.6 GB current
Batching Results
Batch Size TG tok/s Speedup
32.1 1.00×
30.4 0.95×
37.4 1.17×