MiniMax has announced the release of its new video model, Hailuo 2.3, along with Hailuo 2.3 Fast. The official description states that the new version significantly improves body movement, facial expressions, and physical realism, while optimizing motion command response, stylization, and consistency of micro-expressions. The "Fast" version offers lower latency and more controllable costs, suitable for mass production or rapid iteration. The web template page is marked "Introducing Hailuo 2.3 / 2.3 Fast," offering two workflows: text-to-video and image-to-video.
The platform page shows that the Hailuo 2.3 series is now available for trial on the site, offering a daily limit of 4 free generation sessions. Creators can directly access upgraded dance, emoticon, anime/art style, and e-commerce ad templates from the scene library, choosing between realistic quality and generation speed as needed. Third-party platforms have also launched 2.3 Fast documentation and API access points for easy integration into their own workflows.
Frequently Asked Questions
Q: What is the difference between Hailuo 2.3 and 2.3 Fast?
A: 2.3 focuses on cinematic realism and stability; 2.3 Fast prioritizes lower latency and lower cost, suitable for rapid or batch generation.
Q: What are the core improvements in this upgrade?
A: Body movements, facial expressions, and physical realism are more natural, cues/action instructions are more readily followed, and stylization and consistency are enhanced simultaneously.
Q: Does it support text-to-video and image-to-video generation?
A: Yes, both workflows are available on the webpage and third-party platforms.
Q: Is there a free quota?
A: Yes. The website states "4 free generation attempts per day".
Q: Where can I call or integrate this?
A: It can be used directly on the Hailuo web interface; some third-party platforms (such as Replicate, ImagineArt, etc.) provide 2.3/2.3 Fast access and instructions.
