Three new image model families are now live for generation and editing.

Qwen Image 3: bilingual prompt support, automatic prompt expansion, negative prompts, and 1K or 2K output. Text-to-Image Image-to-Image
Muse Image: agentic image generation, precise edits, strong typography, and multi-part prompt following. Text-to-Image Image-to-Image
Nano Banana 2: fast Gemini image generation with resolutions from 0.5K to 4K, optional web grounding, and enhanced reasoning. Text-to-Image Image-to-Image

Prompt

Describe the image to generate or the edit to apply.

Init image

Upload the image to edit.

Aspect ratio

Aspect ratio of the generated image. Select Automatic to let the model decide.

Automatic

Available in Power Mode

Count

Number of results to generate in parallel.

More options

Format

The encoding format of the generated image.

PNG

This is a SFW (Safe For Work) only model.