Tencent Hunyuan

Verdict: Tencent Hunyuan is not a single chatbot but a full-stack multimodal model matrix covering language, vision, 3D, video, and translation, targeting developers, content creators, and enterprise teams for complex creation, reasoning, and agentic tasks. What it does: General models Hy2.0 and Hy3 preview fuse fast/slow thinking and strong agent capabilities, deployed in Tencent Yuanbao, Docs, QQ, and Game for Peace NPCs. The open-source Hy-MT2 translation models run on-device (1.8B at 440 MB), with the 7B variant outperforming some closed-source competitors. The team also open-sourced 3D generation pipelines, the UniRL multimodal RL framework, and detailed Hy3 inference optimizations. Cost & entry: Open-source translation, 3D, and UniRL can be deployed locally for free; cloud API pricing is not yet officially disclosed. Community demos show local deployment on a single consumer GPU. Suited for developers and enterprises exploring multimodal, translation, 3D generation, or agent applications; not ideal for those seeking a simple chat interface or zero-code solutions.