← Back to community benchmarks

Qwen3-8B

M3 Pro (18c) · 36 GB · 4bit · 2026-05-12
Performance
1k
tokens
336.3
PP tok/s
27.4
TG tok/s
3045
TTFT (ms)
5.0
Peak mem (GB)
Hardware
Chip M3 Pro (18c)
Memory 36 GB
GPU Cores 18
Software
oMLX v0.3.9.dev1
macOS macOS 26.2
Context 1,024
Performance by Context Length
Context PP tok/s TG tok/s Peak Mem
1k 336.3 27.4 5.0 GB current
4k 337.3 25.5 5.7 GB view
Batching Results
Batch Size TG tok/s Speedup
27.4 1.00×
55.1 2.01×
78.4 2.86×