模型指南

A model is the AI engine that turns your prompt into an actual image or video. Each model is good at different things, supports different options, and uses a different amount of credits — so picking the right one for what you want to make is the first step toward a good result.

Image models vs. video models

• Image models — Create still images from text or a reference image.
• Video models — Create short videos from text or an image. They support video-specific options like resolution and length.

The Image screen offers image models and the Video screen offers video models, so you only ever choose a model that fits the screen you're on.

Choosing a model

Pick a model from the model selector in the generation input. The same prompt can produce a different style, level of detail, and consistency depending on the model. There's no single right answer — it's worth trying a few and comparing until you get a result you like. Switching models also updates the options available to match that model.

Featured models

Among the curated models, you can use Seedance for video and Nano Banana for images.

• Seedance 2.0 (video) — Supports everything from 480p up to 4K (Ultra HD, 3840×2160). 4K is great for big screens and post-production; it uses more credits, so reach for it when you need it.
• Nano Banana (image) — Used for text- and reference-based image generation.

The available lineup may change over time — you can always check the model selector on the generation screen for what's currently available.

Credits

Generation uses credits, and the amount depends on the model and options (such as resolution and length). See the Credits & pricing guide for buying credits and checking your balance.