Hy3 preview: the first step in rebuilding the Hunyuan large model
Original title:Hy3 preview : 混元大模型重建的第一步
AI summary
On April 23, 2026, Tencent Hunyuan released and open-sourced the Hy3 preview language model. The model is a mixture-of-experts model combining fast and slow thinking, with 295B total parameters, 21B activated parameters and support for a 256K context. It substantially improves complex reasoning, instruction following, in-context learning, coding and agent capabilities, achieving strong competitive results on benchmarks such as SWE-Bench Verified, Terminal-Bench 2.0 and BrowseComp. The model has launched in products including Yuanbao and CodeBuddy, with the API personal Token Plan starting at 28 yuan per month. Weights and code are open-sourced on GitHub, HuggingFace and other platforms, supporting inference frameworks such as vLLM and SGLang.
Key Takeaways
- On April 23,2026,Tencent Hunyuan released and open-sourced the Hy3 preview language model.
- The model is a mixture-of-experts model combining fast and slow thinking,with 295B total…
- It substantially improves complex reasoning,instruction following,in-context learning,coding and…
- 0 and BrowseComp.
- The model has launched in products including Yuanbao and CodeBuddy,with the API personal Token…
Quick Facts
- Published
- Updated
- Source
- 腾讯混元:Research(API)
- Category
- Model Releases
- AI score
- 80
混元重建后的第一个模型 Hy3 preview,快慢思考融合,推理和代码智能体提升明显,开源加低价格,国内开发者做 agent 的可以马上试起来。
Related entities
Source & copyright
This article is based on curated titles and AI summaries from the AIHOT public API, with attribution and links back. Original copyright remains with the original publishers and authors; this site does not mirror full texts.
How to Cite
When citing this article, keep the title, date, and permalink, and credit Orange News as the source.
Orange News: Hy3 preview: the first step in rebuilding the Hunyuan large model. June 24, 2026. https://new.nat6.net/en/news/cmr7m14ks00mhslr94up9u2os/