← Back to community benchmarks

Qwen2.5-7B-Instruct

M5 (10c) · 16 GB · 4bit · 2026-05-18
Performance
4k
tokens
808.7
PP tok/s
28.8
TG tok/s
5065
TTFT (ms)
5.0
Peak mem (GB)
Hardware
Chip M5 (10c)
Memory 16 GB
GPU Cores 10
Software
oMLX v0.3.9.dev2
macOS macOS 26.5
Context 4,096
Performance by Context Length
Context PP tok/s TG tok/s Peak Mem
1k 790.4 30.6 4.5 GB view
4k 808.7 28.8 5.0 GB current
Batching Results
Batch Size TG tok/s Speedup
30.6 1.00×
60.6 1.98×
74.6 2.44×