On Sunday, Tencent’s Hunyuan team unveiled a new artificial intelligence (AI) model capable of generating three-dimensional (3D) environments. Named Hunyuan 3D World Model 1.0, this open-source model allows users to input both text and images. It has the capability to create 360-degree worlds that are not only explorable but also interactive. Once these virtual realms are created, users can navigate them as playable environments. Tencent indicates that the model has potential applications in video game design, virtual reality, and digital content production.
Tencent’s Hunyuan 3D World Model 1.0 Built on Innovative Framework
World generation models represent a specialized field within generative AI, focusing on the creation of geometrically consistent, visually diverse, and efficiently rendered environments. These models can be utilized in various applications, ranging from video game creation to virtual training for robots.
The Hunyuan 3D World Model is available for download from both the company’s GitHub and Hugging Face repositories. Released under a permissive “tencent-hunyuanworld-1.0-community” license, it allows for academic research and commercial applications. Users interested in experimenting with the model without local installations can access Tencent’s platform to start generating worlds.
We’re thrilled to release & open-source Hunyuan3D World Model 1.0! This model enables you to generate immersive, explorable, and interactive 3D worlds from just a sentence or an image.
It’s the industry’s first open-source 3D world generation model, compatible with CG pipelines… pic.twitter.com/CpETdVO7vW
— Hunyuan (@TencentHunyuan) July 27, 2025
According to the project page, current world generation techniques typically concentrate either on video-based or 3D-based methods. While video-based approaches excel in visual diversity, 3D methods ensure consistency and rendering efficiency. However, the Hunyuan team asserts that neither approach provides a comprehensive solution, prompting the development of their unique framework.
This innovative framework employs a semantically layered 3D mesh representation that projects panoramic images into 360-degree world proxies. These proxies can be analyzed and reconstructed to form diverse virtual environments. Researchers used this foundational framework to train the model, enabling it to create coherent, interactive, and explorative 3D worlds.
Hunyuan’s team indicated that the open-source AI model is based on Flux’s image generator, with the potential for adaptation to other models, including Hunyuan Image, Kontext, and Stable Diffusion. The company has not disclosed whether the model will be accessible through application programming interfaces (APIs).
