A research model for 3D reconstruction, and the Tencent licence excludes the UK and EU anyway.
Kenning's take
What it does
HunyuanWorld-Mirror takes images or video, plus optional camera details, and predicts 3D geometry in one pass, including point clouds, depth, surface normals, camera settings and 3D Gaussians. It is a research model, accepted at ICML 2026, and its successor WorldMirror 2.0 ships inside HY-World 2.0. Code and weights are on GitHub and Hugging Face.
- 3D reconstruction from images or video
- Point clouds, depth and normals
- Camera pose and intrinsics prediction
- 3D Gaussian splatting output
- Training and evaluation code released
Who it suits. Researchers and developers working on 3D reconstruction.
Key facts
- Category
- AI models
- Made by
- Tencent
- Based in
- China
- Released
- 22 October 2025
- How to get it
- open weights on Hugging Face and code on GitHub; online demo; Tencent's licence states it does not apply in the UK, EU or South Korea
- Works on
- open-weights
- In the UK
- No
- Free plan
- yes, open weights (outside the UK, EU and South Korea)
FAQ
When was HunyuanWorld-Mirror released?
Tencent released the weights and inference code on 22 October 2025.
What replaced HunyuanWorld-Mirror?
WorldMirror 2.0, part of HY-World 2.0 from April 2026.
Facts from the maker's own pages and the sources listed. Checked 2026-10-11.
Similar tools
ACE-Step 1.5
Open-source music generation model that writes full songs with lyrics and runs on ordinary consumer GPUs.
AlphaGenome Atlas
Google DeepMind database of AI-predicted effects for every possible single-letter change in human DNA.
Amazon Nova 2
Amazon's second family of Nova AI models, covering reasoning, speech conversation and combined text-and-image generation.
Cisco Antares
Cisco's small open-weight models that find where known security vulnerabilities sit in a codebase, running locally.