← Back to community benchmarks

Qwen3.5-397B-A17B

M3 Ultra (80c) · 512 GB · 4bit · 2026-04-06
Performance
4k
tokens
505.8
PP tok/s
34.4
TG tok/s
8098
TTFT (ms)
211.5
Peak mem (GB)
Hardware
Chip M3 Ultra (80c)
Memory 512 GB
GPU Cores 80
Software
oMLX v0.3.1
macOS macOS 15.3
Context 4,096
Performance by Context Length
Context PP tok/s TG tok/s Peak Mem
1k 420.2 36.0 209.7 GB view
4k 505.8 34.4 211.5 GB current
Batching Results
Batch Size TG tok/s Speedup
36.0 1.00×
57.5 1.60×
82.1 2.28×