Images
Create, edit, and enhance images.
How do I generate an AI image?
Turn a prompt and optional references into one or more images.
- Open Create and select Image.
- Use Prompt → Configure → review the compact preview → Generate.
- Auto is the recommended starting point if you're unsure which model to choose.
- Add a plain-language prompt, choose an image count, and optionally add up to 14 reference images.
- Configure is free. It shows the selected model, format, quality, reason, warnings, and expected usage before you generate.
- Generate uses the exact configuration shown and does not silently switch models.
Manual remains available when you want direct control.
- Seedream 5.0 (BytePlus) — Fast, high-res. Supports reference images.
- Seedream 5.0 Pro (BytePlus) — Premium tier. Best reference consistency and image editing.
- GPT Image 2.5 Flare (OpenAI) — Fast GPT Image 2.5 generation with strong prompt following.
- GPT Image 2.5 Sunburst (OpenAI) — GPT Image 2.5 generation tuned for higher visual quality.
- Nano Banana Pro (Google) — Direct Gemini generation. Strong reference and text rendering.
- Nano Banana Flash (Google) — Fast Gemini generation. Strong all-around image creation and editing.
Which image model should I choose?
Compare reference images, output shape, quality, and batch limits before generating.
- Auto is the recommended starting point if you're unsure which model to choose.
- Add a plain-language prompt, choose an image count, and optionally add up to 14 reference images.
- Configure is free. It shows the selected model, format, quality, reason, warnings, and expected usage before you generate.
- Generate uses the exact configuration shown and does not silently switch models.
- Manual remains available when you want direct control.
- In Manual, choose the model whose reference-image limit and output controls match what you are trying to make.
Reference images are optional for all current image models, and the controls below update whenever the selected model changes.
- Seedream 5.0 (BytePlus) — Optional reference images (up to 14); Fixed 2K output; Up to 4 images per request; Negative prompt available; Prompt enhancement available
- Seedream 5.0 Pro (BytePlus) — Optional reference images (up to 10); Fixed 2K output; Up to 4 images per request; Prompt enhancement available
- GPT Image 2.5 Flare (OpenAI) — Optional reference images (up to 10); Aspect ratio: 1:1 / 9:16 / 16:9; Quality: Low, Medium, High; Up to 1 image per request; Prompt enhancement available
- GPT Image 2.5 Sunburst (OpenAI) — Optional reference images (up to 10); Aspect ratio: 1:1 / 9:16 / 16:9; Quality: Low, Medium, High; Up to 1 image per request; Prompt enhancement available
- Nano Banana Pro (Google) — Optional reference images (up to 14); Aspect ratio: 1:1 / 9:16 / 16:9; Up to 1 image per request; Prompt enhancement available
- Nano Banana Flash (Google) — Optional reference images (up to 10); Aspect ratio: 1:1 / 9:16 / 16:9; Up to 1 image per request; Prompt enhancement available
Can I use AI images as reference for video generation?
Carry the look of an image into a new video request.
- Yes. In AI Images, open a completed image and select "Use as Reference" to set it as the reference for your next generation.
- Reference images can also be attached directly in AI Videos to guide the visual style.