Video generation
Generate videos from text or from a starting image with Veo 3.1, Kling, Seedance, MiniMax and 13 more models.
Text or image
| Mode | Use it when |
|---|
| Text to video | You want the model to invent the whole scene from your prompt. |
| Image to video | You already have the first frame, from your library or your device, and want it to move. |
Make a video
- Pick a model: Each model lists what it's good at, the lengths it supports and its price per second.
- Write the prompt: Describe the action and the camera movement:
slow dolly in as the waves crash against the lighthouse. - Set the options: Length, aspect ratio, resolution and, on supported models, audio. The cost updates as you change them.
- Generate: Renders take one to several minutes. You can leave the page or start another video; up to 3 render at once.
System prompt
The system prompt is a style guide that is added to every video you make, for example 35mm film look, warm grade, no on-screen text. Set it once to keep a consistent look across a series. It is saved in this browser.
Pricing
Videos are priced per second, and some models charge more for higher resolutions or audio. The exact cost is shown before you generate. If a render fails, the tokens are refunded automatically.
Your videos
- Your videos lists every render, 10 per page, with its status: Queued, Rendering, Ready or Failed.
- Reuse settings copies a video's model, prompt and options back into the form.
- Download saves the MP4.
- Private / On Explore decides whether a watermarked copy is shown on Explore. New videos are on Explore by default.