At the 2025 Tencent Forum of the World Artificial Intelligence Conference held on July 27, 2025, Tencent officially released version 1.0 of the Hunyuan 3D World Model and announced that the model will be fully open-sourced. This move marks the birth of the industry's first open-source world generation model that supports immersive roaming, interaction, and simulation, opening up new possibilities for fields such as game development, virtual reality (VR), and digital content creation.

The Hunyuan 3D World Model 1.0 integrates panoramic visual generation and hierarchical 3D reconstruction technology, capable of accepting text and images as input to quickly generate high-quality, stylistically diverse navigable 3D scenes. This technological breakthrough greatly simplifies the 3D scene construction process. Tasks that previously required professional modeling teams to complete in weeks can now be accomplished in minutes through simple text instructions or image uploads. For game developers, this model can rapidly generate complete 3D scenes containing elements such as buildings, terrain, and vegetation. The output Mesh files can be directly used for game prototypes or level design, while also supporting adjustments to foreground objects and replacement of sky backgrounds, meeting personalized creative needs. Even users without modeling experience can easily generate 360° immersive visual spaces through the Hunyuan 3D creation engine and seamlessly import them into VR headsets like Vision Pro for an immersive experience.

WeChat Screenshot_20250727111432.png

The core advantage of the Hunyuan 3D World Model 1.0 lies in its innovative "Semantic Hierarchical 3D Scene Representation and Generation Algorithm." This algorithm decomposes complex 3D worlds into different semantic levels, enabling intelligent separation of foreground and background, ground and sky. Not only does it generate visually realistic overall scenes, but it also outputs standardized 3D Mesh assets compatible with mainstream tools such as Unity, Unreal Engine, and Blender, allowing users to independently edit or perform physical simulations on scene elements, achieving a seamless integration of AIGC technology with traditional CG workflows. Compared to globally leading open-source models, the Hunyuan 3D World Model 1.0 has achieved comprehensive superiority in key metrics such as aesthetic quality and instruction following ability in text-to-world and image-to-world generation.

In addition, Tencent Hunyuan has disclosed a series of open-source plans, including edge-side hybrid reasoning language models, multimodal understanding models, and game visual models. Among these, the Hunyuan flagship model TurboS has maintained a rapid iteration of one release per month since its launch at the beginning of the year, showing significant improvements in code generation, scientific capabilities, and complex instruction following, firmly ranking among the top global large model authority rankings. Based on the TurboS foundation, the Hunyuan T1 further enhances reasoning capabilities, demonstrating leading comprehensive abilities domestically. At the same time, the end-to-end speech model Hunyuan Voice and the multimodal understanding model Hunyuan Vision are also built upon the Hunyuan large language model foundation, inheriting excellent understanding, cognition, and reasoning capabilities, achieving remarkable results in the field of multimodal understanding.

Tencent Hunyuan also has comprehensive layout in multimodal generation, with global leading 3D generation capabilities. The latest released Hunyuan 3D2.5 model has significantly improved texture quality, ranking first globally in the evaluation by the Shanghai Artificial Intelligence Laboratory. In addition to the Hunyuan 3D World Model 1.0, Tencent will also open-source a series of small-scale models at the end of the month, including 0.5B, 1.8B, 4B, and 7B hybrid reasoning models, which are lighter and easier to deploy. At the same time, the multimodal understanding model Hunyuan-large-vision and the interactive game video generation framework for game scenarios, Hunyuan GameCraft, will also be open-sourced to the public soon.

Tencent Hunyuan has also opened up a complete set of multimodal generation capabilities and toolset plugins, successively open-sourcing industry-leading text-to-image, video generation, and 3D generation capabilities, providing an open-source base close to commercial model performance, making it convenient for the community to customize based on business and usage scenarios. Currently, the community download count of the Hunyuan 3D series models has exceeded 2.3 million times, becoming one of the most popular 3D open-source models worldwide.

Tencent Hunyuan 3D World Model 1.0:

● Project Homepage: https://3d-models.hunyuan.tencent.com/world/

● Experience Address: https://3d.hunyuan.tencent.com/sceneTo3D

● Hugging Face Model Address: https://huggingface.co/tencent/HunyuanWorld-1

● Github Project Address: https://github.com/Tencent-Hunyuan/HunyuanWorld-1.0