← Back to community benchmarks

Qwen3.5-35B-A3B-Heretic

M2 Ultra (60c) · 64 GB · 4bit · 2026-03-13
Performance
16k
tokens
1,121
PP tok/s
62.3
TG tok/s
14616
TTFT (ms)
19.4
Peak mem (GB)
Hardware
Chip M2 Ultra (60c)
Memory 64 GB
GPU Cores 60
Software
oMLX v0.2.10
macOS macOS 26.3.1
Context 16,384
Performance by Context Length
Context PP tok/s TG tok/s Peak Mem
1k 1,138 77.8 18.6 GB view
4k 1,195 74.9 18.8 GB view
8k 1,174 71.6 19.0 GB view
16k 1,121 62.3 19.4 GB current
32k 1,013 51.2 20.2 GB view
64k 829.1 35.1 21.9 GB view
Batching Results
Batch Size TG tok/s Speedup
77.8 1.00×
130.4 1.68×
261.1 3.36×
523.2 6.72×