Google launches Gemini Omni AI video model with multimodal capabilities

Google unveiled Gemini Omni, its first native multimodal AI video model, at the I/O 2026 developer conference. The model can generate 10‑second videos by combining images, video clips, audio and text, and allows users to edit content through natural‑language prompts. Early access is limited to Google AI Plus, Pro and Ultra subscribers, and the service is offered via the Gemini app, Google Flow and YouTube Shorts. Current limits include a maximum of five source images and a small number of generation attempts per user, with additional output formats such as audio and AI avatars still in testing.

During the same keynote Google announced an “intelligent search box” that expands AI‑driven search features. A follow‑up test of six AI tools, including Google Gemini, examined how well each system cites its sources. The study found mixed results, with most tools providing simple hyperlink citations and a sources tab, while the overall quality of citation practices varied widely.