Qwen3.5-397B-A17B
Qwen3.5-397B-A17B is an open-weight language model from Alibaba Qwen, released on February 16, 2026. It is a mixture-of-experts model with 397B total parameters, of which 17B are active for each token. Alibaba Qwen lists a context window of 262,144 tokens natively, extensible to about 1M. The model accepts text and image input.
- Developer
- Alibaba Qwen
- Released
- February 16, 2026
- Total parameters
- 397B
- Active parameters per token
- 17B
- Architecture
- MoE
- Context window
- 256K
- Input
- Text, image
- License
- Apache 2.0
- Commercial use
- Yes
- Official weights, GB
- 806.8 GB
- Hugging Face repository
- huggingface.co
- Checked on
The weights are released under Apache 2.0, which allows commercial use.
The official weights on Hugging Face take about 807 GB. Running the model needs at least that much memory across GPUs and system RAM, plus room for the context cache; quantized versions need less.
- Total parameters, billions
- 397B
- Active parameters, billions
- 17B
- Context window, tokens
- 262.14K
- License type
- Apache 2.0
- License conditions
- Apache 2.0, with no user or revenue limits.
- Official quantized or alternative versions
- Qwen/Qwen3.5-397B-A17B-FP8; Qwen/Qwen3.5-397B-A17B-GPTQ-Int4
- License text
- huggingface.co
- Sources
Page URL Hugging Face model card https://huggingface.co/Qwen/Qwen3.5-397B-A17B/raw/main/README.md Hugging Face weights index https://huggingface.co/Qwen/Qwen3.5-397B-A17B/raw/main/model.safetensors.index.json GitHub file https://raw.githubusercontent.com/QwenLM/Qwen3.6/main/README.md