← Back to community benchmarks

Qwen3.5-4B

M5 (10c) · 24 GB · 4bit · 2026-03-11
Performance
4k
tokens
428.7
PP tok/s
44.0
TG tok/s
9555
TTFT (ms)
4.2
Peak mem (GB)
Hardware
Chip M5 (10c)
Memory 24 GB
GPU Cores 10
Software
oMLX v0.2.7
macOS macOS 26.3.1
Context 4,096
Performance by Context Length
Context PP tok/s TG tok/s Peak Mem
1k 425.6 45.9 3.9 GB view
4k 428.7 44.0 4.2 GB current
Batching Results
Batch Size TG tok/s Speedup
45.9 1.00×
90.2 1.97×
129.2 2.81×