← Back to community benchmarks

Qwen3.5-397B-A17B

M3 Ultra (80c) · 512 GB · 6bit · 2026-03-12
Performance
1k
tokens
458.1
PP tok/s
29.5
TG tok/s
2235
TTFT (ms)
302.7
Peak mem (GB)
Hardware
Chip M3 Ultra (80c)
Memory 512 GB
GPU Cores 80
Software
oMLX v0.2.7
macOS macOS 26.3.1
Context 1,024
Performance by Context Length
Context PP tok/s TG tok/s Peak Mem
1k 458.1 29.5 302.7 GB current
4k 494.4 28.6 302.9 GB view
8k 491.8 26.1 303.3 GB view
16k 470.4 22.5 304.0 GB view
32k 429.1 17.3 305.5 GB view
64k 348.4 12.1 308.5 GB view
Batching Results
Batch Size TG tok/s Speedup
29.5 1.00×
45.8 1.55×
59.3 2.01×