← Back to community benchmarks

Qwen2.5-3B-Instruct

M4 (10c) · 16 GB · 6bit · 2026-03-10
Performance
1k
tokens
543.9
PP tok/s
37.2
TG tok/s
1883
TTFT (ms)
2.9
Peak mem (GB)
Hardware
Chip M4 (10c)
Memory 16 GB
GPU Cores 10
Software
oMLX v0.2.6
macOS macOS 26.0.1
Context 1,024
Batching Results
Batch Size TG tok/s Speedup
1× 37.2 1.00×
2× 68.6 1.84×
4× 73.7 1.98×
8× 75.7 2.03×