← Back to community benchmarks

Qwen3.5-35B-A3B

M5 Pro (20c) · 64 GB · 4bit · 2026-03-28
Performance
64k
tokens
1,279
PP tok/s
40.5
TG tok/s
51261
TTFT (ms)
26.0
Peak mem (GB)
Hardware
Chip M5 Pro (20c)
Memory 64 GB
GPU Cores 20
Software
oMLX v0.2.24
macOS macOS 26.4
Context 65,536
Performance by Context Length
Context PP tok/s TG tok/s Peak Mem
1k 1,964 74.0 20.1 GB view
4k 2,377 73.0 21.0 GB view
8k 2,331 69.4 21.4 GB view
16k 2,136 63.0 22.0 GB view
32k 1,764 53.4 23.4 GB view
64k 1,279 40.5 26.0 GB current
Batching Results
Batch Size TG tok/s Speedup
74.0 1.00×
120.9 1.63×
167.9 2.27×