Microsoft has added a new AI video generator to its Bing mobile app that’s built on OpenAI’s Sora text-to-video model. The Bing Video Creator announced on Monday provides a free way to generate short clips with Sora, which is normally locked behind ChatGPT subscriptions starting at $20 per month for Plus users.
“Bing Video Creator represents our efforts to democratize the power of AI video generation,” Microsoft said in its announcement. “We believe creativity should be effortless and accessible to help you satisfy your answer-seeking process. We’re excited to empower anyone to turn their words into wonder through an AI-generated video.”
The Video Creator is now rolling out globally (excluding China and Russia) to the Bing Search apps for Android and iPhone, and Microsoft says desktop and Copilot Search support are also “coming soon.” The video generator can be accessed via the menu at the bottom right corner of the Bing app, or by adding a description of the clip you want to make directly to the Bing search bar.
Microsoft says that users can queue up to three video generations at a time and will receive a notification when they’re ready. You can choose between a “Standard” generation speed, which is free for all users, or a “Fast” option that works “in seconds,” according to Microsoft. 10 Fast generations are provided for free, after which users will need to use the Standard speed or redeem 100 Microsoft Rewards points for each Fast generation.
Videos are five seconds long, in a 9:16 vertical format, and support for 16:9 format video creation will be added at a later date. Videos will be stored on the Bing app for up to 90 days and can be downloaded or shared with other platforms.
The example videos that Microsoft has released are a far cry from the quality showcased by other AI models like Google’s Veo 3. The movements in the otter chef and sweeping hamster demos show unnatural clipping, for example, and the characters themselves look a little cartoony at times. Still, judging by our time experimenting with Sora after it launched in December 2024, these are a better representation of what the video model is capable of than the polished (and presumably cherry-picked) clips that OpenAI released when Sora was being teased.
