← Back to community benchmarks

Qwen3.5-397B-A17B

M3 Ultra (80c) · 512 GB · 4bit · 2026-03-16
Performance
4k
tokens
514.1
PP tok/s
33.9
TG tok/s
7968
TTFT (ms)
210.7
Peak mem (GB)
Hardware
Chip M3 Ultra (80c)
Memory 512 GB
GPU Cores 80
Software
oMLX v0.2.10
macOS macOS 15.7.4
Context 4,096
Performance by Context Length
Context PP tok/s TG tok/s Peak Mem
1k 481.4 35.3 210.4 GB view
4k 514.1 33.9 210.7 GB current
8k 510.9 32.2 211.0 GB view
16k 489.3 25.6 211.8 GB view
32k 440.8 18.5 213.2 GB view
Batching Results
Batch Size TG tok/s Speedup
35.3 1.00×
55.8 1.58×
82.4 2.33×
115.5 3.27×