Video models
| Model | Input | Best for |
|---|---|---|
| Sora 2 Pro | Text | Cinematic narrative shots |
| Kling 2.6 | Text / Image | Controlled camera movement |
| Wan 2.6 | Image | Animating an existing still with audio |
| Kling Avatar + ElevenLabs | Image + Script | Lip-synced spokesperson clips |
| Topic to Video | Topic | A complete video from one idea |
Image models
| Model | Best for |
|---|---|
| Seedream 4.5 | Photorealistic detail and product stills |
| Nano Banana Pro | Fast edits, variations and iterations |
How to choose in three questions
- Do you already have the visual? If yes, use an image-to-video model so the subject stays exactly as it is. If no, start with a text-to-video or image model.
- Does someone need to speak? If yes, use the avatar model rather than trying to prompt speech out of a general video model.
- Do you need one shot or a whole video? One shot is a single model call; a whole video is Topic to Video.
Specialised tools
- AI birthday video maker — personalized birthday videos from one photo.
- Viral video presets — ready-made formats built on these models.
Costs
Each model has its own credit cost, listed on the pricing page. Preview in standard quality first and reserve high quality renders for the version you intend to publish.