← Back to community benchmarks

Qwen3.5-35B-A3B

M5 Pro (20c) · 64 GB · 4bit · 2026-03-18
Performance
16k
tokens
1,961
PP tok/s
76.0
TG tok/s
8354
TTFT (ms)
22.0
Peak mem (GB)
Hardware
Chip M5 Pro (20c)
Memory 64 GB
GPU Cores 20
Software
oMLX v0.2.18
macOS macOS 26.3.1
Context 16,384
Performance by Context Length
Context PP tok/s TG tok/s Peak Mem
1k 1,915 95.4 20.1 GB view
4k 2,298 90.1 21.0 GB view
8k 2,242 85.4 21.4 GB view
16k 1,961 76.0 22.0 GB current
32k 1,625 60.3 23.4 GB view
Batching Results
Batch Size TG tok/s Speedup
95.4 1.00×
148.2 1.55×
199.4 2.09×