Bedu allegedly released a new artificial intelligence (AI) video generation model on Wednesday. According to the report, the Mystermer AI model can also connect Chinese audio in the videos produced, which can make Google’s second model after VEO 3. Tech Dev has claimed that it is the first AI model in the world to support local Chinese audio generation. Along with the introduction of a large language model (LLM), the company allegedly launched a new video content created platform called Huxiang. In particular, neither Messenger nor Hexiang is currently available outside China.
Bedo’s Mesterimer may allegedly produce Chinese audio
The world of the AI video generation model has been significantly developed in the past two years. We have moved from models that are struggling to produce people in LLM with a fixed number of fingers that can now properly present realistic physics and movements. However, most of the AI players in one area refrained from entering were videos that also supported audio locally.
In Google I/O 2025, Tech Dev became the first company to offer this capacity with VEO 3, which immediately. It became the talk of the city, which left its greatest rival, Openi’s Surah. Mountain View -based Tech Dev recently increased VEO 3 in all 154 countries where the Gemini app is available, which highlighted the company’s aggressive pressure for the device.
However, according to a Tech in Asia Report (AI Base), Chinese Tech Dev Bedo has also enrolled in the race with its Masteme AI model. It is said to produce videos with Chinese audio, and the only model with the ability to do so. In particular, VEO 3 can only produce audio in English.
Mystermers are reportedly not only in dialogue that is compatible with videos, but they can also include sound effects and acne noise in the videos. It is said that Bedu has claimed that the model scored 89.38 % on the VBench I2V benchmark, which is at the top. Tech giant is pulling LLM as a content creation tool for users.
Along with the AI model, Bedu has also launched a new video content platform called Hexiang. The hexiang is said to work at the front end of the AI model, where users can distribute indicators and produce videos. The report says that this platform supports 10 seconds long video breeds in 1080p resolution. In comparison, VEO 3 can produce just eight seconds long videos. The default aspect of the video is not more specified than the proportion, and if users can produce videos in different aspects.
