← Back to community benchmarks

Qwen3-4B-Instruct-2507

M4 Pro (20c) · 24 GB · 4bit · 2026-03-10
Performance
1k
tokens
720.2
PP tok/s
74.6
TG tok/s
1422
TTFT (ms)
2.9
Peak mem (GB)
Hardware
Chip M4 Pro (20c)
Memory 24 GB
GPU Cores 20
Software
oMLX v0.2.7
macOS macOS 15.5
Context 1,024
Batching Results
Batch Size TG tok/s Speedup
1× 74.6 1.00×
2× 133.3 1.79×
4× 162.8 2.18×
8× 174.0 2.33×