Meta unveils muse video model with 10-second AI video generation capabilities

A testingcatalog report revealed that Meta is set to launch its Muse Video model, which has already entered a beta testing phase with select partners. The model can generate video clips up to 10 seconds in length. According to the media's tests, Muse Video demonstrates strong potential in terms of detail rendering, understanding of scene and object relationships, and temporal stability across consecutive frames. However, Meta acknowledged that the model still faces challenges in synchronizing audio with visual elements and accurately representing fast-moving physics. The current samples are limited to 10-second clips, which are insufficient to assess the model's performance in longer narratives, complex character consistency, or high-intensity action sequences. The example provided includes a static shot of a glass of red wine tipping over on a white marble counter, spilling toward the camera.