← Back to community benchmarks

Qwen3.6-35B-A3B-uncensored-heretic-Native-MTP-Preserved

M5 Max (40c) · 128 GB · fp16 · 2026-07-25
Performance
64k
tokens
2,108
PP tok/s
54.9
TG tok/s
31093
TTFT (ms)
71.5
Peak mem (GB)
Hardware
Chip M5 Max (40c)
Memory 128 GB
GPU Cores 40
Software
oMLX v0.5.3
macOS macOS 27.0
Context 65,536
Performance by Context Length
Context PP tok/s TG tok/s Peak Mem
1k 3,060 70.6 65.8 GB view
4k 3,353 68.6 66.6 GB view
8k 3,440 68.2 66.9 GB view
16k 3,188 66.5 67.6 GB view
32k 2,851 61.1 68.9 GB view
64k 2,108 54.9 71.5 GB current
128k 1,237 45.0 76.9 GB view
195k 835.4 38.6 82.2 GB view
Batching Results
Batch Size TG tok/s Speedup
70.6 1.00×
69.0 0.98×
118.9 1.68×