Tencent Hunyuan: AI Models, API Search and Web Access
Table of Contents
Tencent Hunyuan is Tencent’s family of AI models for text, images, video and 3D. A model has no web access of its own. The Hunyuan API on Tencent Cloud has an optional search feature with citation links. Tencent does not document a crawler or user agent for it, so a site owner has no Hunyuan rule to add.
The Hunyuan models
Tencent publishes Hunyuan models in its Tencent-Hunyuan GitHub organization. Some models in that organization:
| Model | Type | License |
|---|---|---|
| Hunyuan-A13B | Language model, open-sourced June 27, 2025 | Tencent Hunyuan Community License |
| Hy3 | Language and agent model from the Tencent Hy Team | Apache License 2.0 |
| HunyuanVideo | Video generation | See the repository |
| Hunyuan3D-2.1 | 3D assets from images | See the repository |
Hunyuan-A13B has 80 billion parameters, and 13 billion of them are active. Active parameters are the part of the model that works on each token. Its license says that it does not apply in the European Union, the United Kingdom and South Korea. Hy3 has 295 billion parameters, and 21 billion are active.
Search in the Hunyuan API
Developers can also call Hunyuan models through Tencent Cloud. The OpenAI-compatible API documentation (in Chinese) lists these search options:
| Parameter | What it does |
|---|---|
enable_enhancement |
Turns on extra features such as search. The default is off. |
force_search_enhancement |
Makes the model always use AI search |
search_info |
Returns the search results when a search runs |
citation |
Adds reference marks to the answer. Each mark points to a link in the search results. |
The documentation does not say which search engine or crawler supplies the results. Search is off by default, so an answer uses web results only when the developer turns search on.
Crawlers and blocking
Tencent does not document a user agent, a robots.txt token or IP ranges for Hunyuan. If an app that uses Hunyuan reads web pages directly, the app sends those requests. For bots that read the web for AI, see AI Crawler Bots: Training, Search and User Fetchers.