← Back to community benchmarks

Qwen3-4B-Instruct-25075bit

M3 Pro (18c) · 36 GB · 5bit · 2026-03-10
Performance
1k
tokens
575.7
PP tok/s
38.0
TG tok/s
1779
TTFT (ms)
3.4
Peak mem (GB)
Hardware
Chip M3 Pro (18c)
Memory 36 GB
GPU Cores 18
Software
oMLX v0.2.7
macOS macOS 26.3.1
Context 1,024
Performance by Context Length
Context PP tok/s TG tok/s Peak Mem
1k 575.7 38.0 3.4 GB current
4k 529.7 33.5 3.9 GB view
Batching Results
Batch Size TG tok/s Speedup
1× 38.0 1.00×
2× 74.4 1.96×
4× 84.3 2.22×