← Back to community benchmarks

Qwen3.5-35B-A3B9bit

M3 Ultra (60c) · 96 GB · 8bit · 2026-03-09
Performance
1k
tokens
1,268
PP tok/s
74.5
TG tok/s
807
TTFT (ms)
37.4
Peak mem (GB)
Hardware
Chip M3 Ultra (60c)
Memory 96 GB
GPU Cores 60
Software
oMLX v0.2.6
macOS macOS 26.3.1
Context 1,024
Performance by Context Length
Context PP tok/s TG tok/s Peak Mem
1k 1,268 74.5 37.4 GB current
4k 1,877 72.8 37.5 GB view
8k 1,940 69.3 37.7 GB view
16k 1,849 64.1 38.1 GB view
32k 1,620 51.3 39.0 GB view
Batching Results
Batch Size TG tok/s Speedup
74.5 1.00×
127.5 1.71×
199.8 2.68×