腾讯混元组织架构再升级，合并大语言模型和多模态团队，加速模型研发
Merged summary
TL;DR - Tencent merged its Hunyuan large-language and multimodal model teams into a unified foundation-model division led by chief AI scientist Yao Shunyu. The reorganization aims to accelerate development, improve collaboration, and advance unified multimodal capabilities.
- Hy3 reportedly matches flagship models with 2–5 times more parameters and has been integrated into several Tencent products and API platforms.
- Hy3 usage grew more than 68-fold over Hy2 within a week of launch; 60% of WorkBuddy users selecting a model chose Hy3.
- Hunyuan’s portfolio spans language, image, video, speech, and 3D generation.
- The open-source Hy World 2.0 generates and reconstructs 3D environments from text, images, or video, while Hunyuan 3D models have surpassed 3 million community downloads.
Sources (1)
腾讯混元组织架构再升级，合并大语言模型和多模态团队，加速模型研发
TL;DR - Tencent merged its Hunyuan large-language and multimodal model teams into a unified foundation-model division led by chief AI scientist Yao Shunyu. The reorganization aims to accelerate development, improve collaboration, and advance unified multimodal capabilities.
- Hy3 reportedly matches flagship models with 2–5 times more parameters and has been integrated into several Tencent products and API platforms.
- Hy3 usage grew more than 68-fold over Hy2 within a week of launch; 60% of WorkBuddy users selecting a model chose Hy3.
- Hunyuan’s portfolio spans language, image, video, speech, and 3D generation.
- The open-source Hy World 2.0 generates and reconstructs 3D environments from text, images, or video, while Hunyuan 3D models have surpassed 3 million community downloads.