Tencent launches Hunyuan 3D World Model to power game development and virtual experiences
Tencent unveils an open-source AI model that can build explorable 3D worlds from text and images.
- Jul 29, 2025,
- Updated Jul 29, 2025 12:52 PM IST
Tencent’s Hunyuan team has released a new open-source artificial intelligence model designed to generate fully explorable and interactive 3D environments. Called the Hunyuan 3D World Model 1.0, the system allows users to create immersive virtual worlds from either text or image prompts, with potential applications across gaming, virtual reality and digital content creation.
The model was officially launched on Sunday and is now available for both academic and commercial use through Tencent’s GitHub and Hugging Face repositories. Those who prefer not to run it locally can also test the model directly through Tencent’s online platform.
Unlike earlier methods that focused solely on video-based or 3D-based world generation, Tencent’s approach blends both visual diversity and geometric consistency. “While the former provides strong visual diversity, the latter provides 3D consistency and efficiency in rendering. However, none of these approaches offers both,” the Hunyuan team stated on the project page. Their solution introduces a new framework built on a semantically layered 3D mesh representation.
This structure allows panoramic images to act as proxies for 360-degree virtual environments, which are then decomposed and reconstructed into detailed, interactive spaces. The final output can serve as playable assets that users can navigate in real time.
The model is based on Flux’s image generator but can also be adapted for use with other tools such as Hunyuan Image, Kontext and Stable Diffusion. Although the company has not confirmed API access at this stage, the permissive licence allows wide use, including for commercial development.
By offering both high-quality rendering and real-time interactivity, Tencent’s Hunyuan 3D World Model enters a growing niche within generative AI: the automated creation of explorable digital environments. These tools are increasingly used to design game levels, build virtual worlds for training AI agents and develop immersive content for the metaverse.
For Unparalleled coverage of India's Businesses and Economy – Subscribe to Business Today Magazine
Tencent’s Hunyuan team has released a new open-source artificial intelligence model designed to generate fully explorable and interactive 3D environments. Called the Hunyuan 3D World Model 1.0, the system allows users to create immersive virtual worlds from either text or image prompts, with potential applications across gaming, virtual reality and digital content creation.
The model was officially launched on Sunday and is now available for both academic and commercial use through Tencent’s GitHub and Hugging Face repositories. Those who prefer not to run it locally can also test the model directly through Tencent’s online platform.
Unlike earlier methods that focused solely on video-based or 3D-based world generation, Tencent’s approach blends both visual diversity and geometric consistency. “While the former provides strong visual diversity, the latter provides 3D consistency and efficiency in rendering. However, none of these approaches offers both,” the Hunyuan team stated on the project page. Their solution introduces a new framework built on a semantically layered 3D mesh representation.
This structure allows panoramic images to act as proxies for 360-degree virtual environments, which are then decomposed and reconstructed into detailed, interactive spaces. The final output can serve as playable assets that users can navigate in real time.
The model is based on Flux’s image generator but can also be adapted for use with other tools such as Hunyuan Image, Kontext and Stable Diffusion. Although the company has not confirmed API access at this stage, the permissive licence allows wide use, including for commercial development.
By offering both high-quality rendering and real-time interactivity, Tencent’s Hunyuan 3D World Model enters a growing niche within generative AI: the automated creation of explorable digital environments. These tools are increasingly used to design game levels, build virtual worlds for training AI agents and develop immersive content for the metaverse.
For Unparalleled coverage of India's Businesses and Economy – Subscribe to Business Today Magazine