← Back to community benchmarks

Qwen3.5-4B

M5 (10c) · 32 GB · 4bit · 2026-07-14
Performance
32k
tokens
945.0
PP tok/s
32.3
TG tok/s
34676
TTFT (ms)
6.5
Peak mem (GB)
Hardware
Chip M5 (10c)
Memory 32 GB
GPU Cores 10
Software
oMLX v0.4.4
macOS macOS 26.5.1
Context 32,768
Performance by Context Length
Context PP tok/s TG tok/s Peak Mem
1k 1,110 46.7 3.3 GB view
4k 1,268 44.9 3.9 GB view
32k 945.0 32.3 6.5 GB current
64k 681.9 24.0 9.6 GB view
Batching Results
Batch Size TG tok/s Speedup
46.7 1.00×
91.7 1.96×