← Back to community benchmarks

Qwen3.5-4B

M3 (10c) · 16 GB · 4bit · 2026-03-10
Performance
1k
tokens
314.5
PP tok/s
30.9
TG tok/s
3256
TTFT (ms)
3.9
Peak mem (GB)
Hardware
Chip M3 (10c)
Memory 16 GB
GPU Cores 10
Software
oMLX v0.2.6
macOS macOS 26.2
Context 1,024
Performance by Context Length
Context PP tok/s TG tok/s Peak Mem
1k 314.5 30.9 3.9 GB current
4k 324.7 27.0 4.2 GB view
Batching Results
Batch Size TG tok/s Speedup
30.9 1.00×
56.1 1.82×
76.0 2.46×
78.8 2.55×