← Back to community benchmarks

Qwen3.6-35B-A3B

M3 Ultra (80c) · 512 GB · 4bit · 2026-05-12
Performance
32k
tokens
2,225
PP tok/s
82.8
TG tok/s
14726
TTFT (ms)
255.0
Peak mem (GB)
Hardware
Chip M3 Ultra (80c)
Memory 512 GB
GPU Cores 80
Software
oMLX v0.3.9.dev1
macOS macOS 26.4
Context 32,768
Performance by Context Length
Context PP tok/s TG tok/s Peak Mem
1k 1,771 99.3 251.9 GB view
4k 2,618 99.2 252.7 GB view
8k 2,663 93.6 253.0 GB view
16k 2,582 89.8 253.7 GB view
32k 2,225 82.8 255.0 GB current
64k 1,674 71.4 257.7 GB view
128k 1,040 56.8 263.1 GB view
Batching Results
Batch Size TG tok/s Speedup
99.3 1.00×
179.5 1.81×