← Back to community benchmarks

Qwen3.6-35B-A3B-oQ4-fp16-mtp

M1 Max (24c) · 32 GB · 4bit · 2026-06-07
Performance
32k
tokens
638.9
PP tok/s
41.2
TG tok/s
51285
TTFT (ms)
23.5
Peak mem (GB)
Hardware
Chip M1 Max (24c)
Memory 32 GB
GPU Cores 24
Software
oMLX v0.4.1
macOS macOS 26.3.1
Context 32,768
Performance by Context Length
Context PP tok/s TG tok/s Peak Mem
1k 592.2 56.3 20.3 GB view
4k 741.3 53.4 20.5 GB view
32k 638.9 41.2 23.5 GB current
64k 463.5 33.4 25.2 GB view
Batching Results
Batch Size TG tok/s Speedup
56.3 1.00×
70.5 1.25×
91.8 1.63×