← Back to community benchmarks

Qwen3.5-4B-mxfp4

M4 Pro (20c) · 24 GB · 4bit · 2026-03-14
Performance
8k
tokens
557.6
PP tok/s
54.0
TG tok/s
14692
TTFT (ms)
6.1
Peak mem (GB)
Hardware
Chip M4 Pro (20c)
Memory 24 GB
GPU Cores 20
Software
oMLX v0.2.11
macOS macOS 26.3.1
Context 8,192
Performance by Context Length
Context PP tok/s TG tok/s Peak Mem
1k 674.6 70.9 5.0 GB view
4k 579.6 64.4 5.4 GB view
8k 557.6 54.0 6.1 GB current
Batching Results
Batch Size TG tok/s Speedup
70.9 1.00×
112.1 1.58×
142.4 2.01×
136.8 1.93×