What makes the clips
The AI Vids models: Preview and Long take simulated in your browser, Veo 3.1 and Kling 2.1 connected in code and ready where their keys are set.
| Model | Runs | Clips | State |
|---|---|---|---|
| Preview | In your browser | 2, 4, 6 or 8 s, start and end pictures | Ready, simulated |
| Long take | In your browser | Up to 60 s, start picture only | Ready, simulated |
| Veo 3.1 | On the server | Up to 8 s with sound, from words or first and last frame | Connected in code; ready where its key is set |
| Kling 2.1 | On the server | 10 s from a start picture, optionally an end picture | Connected in code; ready where its key is set |
The Previews paint real clips with WebCodecs from your prompt and a seed. They are not video models. The models list in the app says what is ready where you are. Your prompt and start picture go to a model's provider when one is used.