← Back to community benchmarks

Qwen3.6-35B-A3B-oQ4

M4 Max (40c) · 48 GB · 4bit · 2026-04-17
Performance
8k
tokens
1,727
PP tok/s
105.6
TG tok/s
4745
TTFT (ms)
21.1
Peak mem (GB)
Hardware
Chip M4 Max (40c)
Memory 48 GB
GPU Cores 40
Software
oMLX v0.3.6
macOS macOS 26.5
Context 8,192
Performance by Context Length
Context PP tok/s TG tok/s Peak Mem
1k 1,278 114.0 19.9 GB view
4k 1,735 110.3 20.7 GB view
8k 1,727 105.6 21.1 GB current
16k 1,574 96.0 21.7 GB view
Batching Results
Batch Size TG tok/s Speedup
114.0 1.00×
181.1 1.59×
243.8 2.14×
303.3 2.66×