Repository Public
1The Gemini API offers two models for generating video,2[Gemini Omni Flash](https://ai.google.dev/gemini-api/docs/omni) and [Veo](https://ai.google.dev/gemini-api/docs/veo).3Each are designed for different workflows.45Use Gemini Omni Flash as your default model for video generation. It provides6superior video coherence, multi-input reasoning (supporting text, images, audio,7and video inputs simultaneously), character consistency, factual accuracy, and8multi-turn conversational editing (e.g., element replacement or perspective9changes). Use Veo 3.1 for specific10capabilities like scene extension, last-frame control, or integration with11legacy pipelines are required.1213## Gemini Omni Flash1415Gemini Omni Flash is a fast, multimodal model for video generation and16conversational video editing. It excels at quickly turning text prompts and17images into short videos, and lets you refine results across multiple turns18using the Interactions API.1920[Get started with Gemini Omni Flash →](https://ai.google.dev/gemini-api/docs/omni)2122## Veo 3.12324Veo 3.1 is a model for generating video with native audio. It supports25features like video extension, frame-specific generation, and image-based26direction through the `generateContent` API.2728[Get started with Veo 3.1 →](https://ai.google.dev/gemini-api/docs/veo)2930## Video understanding3132If you need to ingest and analyze existing video content rather than generate33new video, see the [Video understanding guide](https://ai.google.dev/gemini-api/docs/video-understanding).