AI BriefingSand AI Releases MAGI-2 Preview
·2026.08.15 05:50
Key point
Sand AI has released the MAGI-2 preview, a 114B parameter audio-visual MoE model.
Details
Sand AI has released the MAGI-2 Preview for efficient scaling of video generation models.
This model utilizes a 114B AV MoE (Audio-Visual Mixture of Experts) architecture to process audio and video data integratively, supporting high-quality video generation.
Key features are as follows:
- Efficient Scaling: Designed with a MoE structure for large-scale parameters to enhance computational efficiency while securing strong generation performance.
- Audio-Visual Integration: Aims for more sophisticated and immersive video generation by combining visual and audio information.
Related models and code are currently available via Hugging Face and GitHub.
This summary was generated automatically by AI. Check the original for the author's claims and context. Copyright belongs to the original author.
Our guide explains how the AI works. Report summary errors, attribution issues, or removal requests via Contact.
Related
JD Open Source Releases JoyAI-Echo: 5-Minute Video Generation and Omnmodal World Model Support
Step 5 Preview Released: 600B Parameter MoE Model Highlights Agentic Tasks and Financial Analysis PerformanceOpenComputer Hosts 'Ship Video' Agent Using Claude Opus 5.5 to Generate Explainer Videos from Code