MiniMax, a Chinese artificial intelligence (AI) unicorn startup backed by Alibaba and Tencent, released a text-to-video generation model last week. Dubbed Video-01, the platform is available to the public even outside of China. It can generate six-second-long videos in a wide range of styles. Video-01 joins the company's existing AI offerings such as a music generator, generative speech, and language translation. The AI video generator will compete with similar tools in the market such as Runway AI, Pika 1.0, and soon-to-be-released Sora by OpenAI.
MiniMax Video-01 Launched
The AI video generator was launched by MiniMax Founder and Chief Executive Officer Yan Junjie at the company's first developer conference in Shanghai, according to a South China Morning Post report. The founder reportedly highlighted that Video-01 is the first iteration of a text-to-video generation model and future updates will allow users to generate videos from image inputs. Video editing capabilities are also said to be added via future updates.
Gadgets 360 staff members were able to access the platform and generate videos by writing a text prompt in English. The MiniMax Video-01 can generate six-second-long videos in 1280 x 720 pixels resolution at 25 frames per second. The tool can generate videos in a wide range of styles and perspectives.
Enthusiasts do not need to register to use the platform and can directly generate videos. Video-01 took about five minutes to generate videos in our testing, but this could be due to a high number of users accessing the platform simultaneously. There is a limit for free users, however, the limit is currently not known.
A demo of the generation capability of the MiniMax Video-01 was posted on X (formerly known as Twitter). While the demo showcases high-quality rendering, we found the video generation to be of the same level as Runway and Pika 1.0. The details about the data procurement to train the AI model or the user privacy for the generated videos are not known.
During our testing, we also found that Video-01 was generating videos with influential people, which can contribute to deepfakes. Whether this would be addressed in future updates is not known.
