← Back to community benchmarks

Qwen3.6-35B-A3BMTP

M5 Pro (20c) · 48 GB · 4bit · 2026-05-18
Performance
16k
tokens
2,159
PP tok/s
87.6
TG tok/s
7590
TTFT (ms)
21.7
Peak mem (GB)
Hardware
Chip M5 Pro (20c)
Memory 48 GB
GPU Cores 20
Software
oMLX v0.3.9.dev2
macOS macOS 26.4.1
Context 16,384
Performance by Context Length
Context PP tok/s TG tok/s Peak Mem
1k 1,479 101.8 19.7 GB view
4k 2,211 96.0 20.5 GB view
8k 2,242 93.6 21.0 GB view
16k 2,159 87.6 21.7 GB current
32k 1,837 73.8 23.2 GB view
64k 1,350 59.4 26.2 GB view
Batching Results
Batch Size TG tok/s Speedup
101.8 1.00×
102.9 1.01×
351.9 3.46×