Hunyuan's smartest model yet: Tencent releases and open-sources Hy3 preview language model
AI summary
Tencent released and open-sourced the Hy3 preview language model, the smartest in its Hunyuan series to date. The model uses a mixture-of-experts architecture integrating fast and slow thinking, with 295 billion total parameters and support for up to 256K context length. As the first model after infrastructure reconstruction, it achieves major improvements in complex reasoning, instruction following, coding, and other capabilities. It has been applied in internal products such as Yuanbao, QQ, and Tencent Docs, and supports open-source agents such as OpenClaw. Tencent Cloud also launched an API service, with input priced as low as RMB 1.2 per million tokens, and offers a personal Token Plan starting at RMB 28 per month for 35 million tokens.
Key Takeaways
- Tencent released and open-sourced the Hy3 preview language model,the smartest in its Hunyuan…
- The model uses a mixture-of-experts architecture integrating fast and slow thinking,with 295…
- As the first model after infrastructure reconstruction,it achieves major improvements in complex…
- It has been applied in internal products such as Yuanbao,QQ,and Tencent Docs,and supports…
- Tencent Cloud also launched an API service,with input priced as low as RMB 1.
Quick Facts
- Published
- Updated
- Source
- IT之家·人工智能
- Category
- Model Releases
- AI score
- 70
腾讯混元重建训练体系后的首个模型,295B总参但推理时只激活21B,性价比思路很实际,定价比同体量模型便宜一截,开发者值得试试。
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: Hunyuan's smartest model yet: Tencent releases and open-sources Hy3 preview language model. April 23, 2026. https://new.nat6.net/en/news/cmob79ikr06cksl1y74eyvgo9/