← Back to community benchmarks

Qwen3.5-122B-A10B

M3 Ultra (80c) · 512 GB · 4bit · 2026-03-19
Performance
128k
tokens
384.0
PP tok/s
18.7
TG tok/s
341308
TTFT (ms)
86.7
Peak mem (GB)
Hardware
Chip M3 Ultra (80c)
Memory 512 GB
GPU Cores 80
Software
oMLX v0.2.18
macOS macOS 26.3.1
Context 131,072
Performance by Context Length
Context PP tok/s TG tok/s Peak Mem
1k 920.7 51.9 66.4 GB view
8k 964.7 45.7 68.5 GB view
32k 784.2 35.7 72.1 GB view
128k 384.0 18.7 86.7 GB current
Batching Results
Batch Size TG tok/s Speedup
51.9 1.00×
88.4 1.70×
135.9 2.62×
194.0 3.74×