← Back to community benchmarks

Qwen3.5-35B-A3B6.5bit

M3 Ultra (80c) · 512 GB · 6bit · 2026-03-10
Performance
1k
tokens
1,401
PP tok/s
73.8
TG tok/s
731
TTFT (ms)
27.3
Peak mem (GB)
Hardware
Chip M3 Ultra (80c)
Memory 512 GB
GPU Cores 80
Software
oMLX v0.2.7
macOS macOS 26.3
Context 1,024
Performance by Context Length
Context PP tok/s TG tok/s Peak Mem
1k 1,401 73.8 27.3 GB current
4k 2,206 70.9 27.4 GB view
8k 2,301 66.8 27.6 GB view
Batching Results
Batch Size TG tok/s Speedup
73.8 1.00×
118.8 1.61×
195.3 2.65×
397.9 5.39×