Gemini Omni excels at cross-modal reasoning

Gemini Omni, announced at Google I/O 2026, turns sketches and text prompts into finished videos in under sixty seconds. It handles text, speech, images, and video inputs, leveraging physics knowledge for realistic generation. Users mainly complain about compute-based quotas draining faster than expected.