Alibaba

Qwen3.5 397B-A17B

MoE

403,4B

Parameters · 17B active

231,2 Go

Q4_K_M file

≥ 232,7 Go

Recommended VRAM

262 144

Max context

Not measured yet

No card has run the TokenBench protocol on this model yet. The current protocol covers a selection of models; benchmarking any catalog model from the app is coming soon.

Measure my card →

Which cards should it fit on?

Cards on the site whose video memory covers the recommended 232,7 Go (file + 4,000-token context). An estimate, not a measurement. Click a card for the detailed estimate.

Technical sheet

Published: 2026-02-16
Architecture: 60 layers · 2 KV heads · dimension 256
Attention: Hybrid: 15 of 60 layers use standard attention, the others a fixed-memory mechanism (Mamba, linear attention) — a very light cache.
Context cache (f16): ≈ 29 MB per 1,000 tokens · 0,9 Go for 32,768 tokens
GGUF file: bartowski/Qwen_Qwen3.5-397B-A17B-GGUF · 7 files Download (231,2 Go)
Official repo: Qwen/Qwen3.5-397B-A17B

Public Hugging Face data (official config.json, GGUF repo). The context cache only counts full-attention layers.