CASE / 01
An actual anime illustration
A seaside platform, an adult traveler and sunset. See the actual landscape output, full prompt, aspect ratio and charge.
Learn more01 / CREATE
AI image generation and image API.
Use Nano Banana, GPT Image and other models in the browser or connect the image API to your app. Choose the version, resolution and ratio for the model, then download the image and check the charge.
CASE / 02
An actual e-commerce poster
Fictional skincare products, a Chinese headline and texture details. See how a complete promotional poster was generated from a prompt.
Learn moreSTART / 03
Your first image
Choose a model, resolution and aspect ratio, save the result, then check the actual charge in your logs.
Learn moreNano Banana 2, 2.1 and Pro: versions and resolution
The workspace groups Nano Banana into one model entry. Choose 2, 2.1 or Pro with the version control. Each offers native 1K, 2K and 4K tiers; API requests use the separate model IDs below. Check the version and current quote before choosing a resolution for your task.
| Workspace version | API model ID | Resolution tiers |
|---|---|---|
| Nano Banana 2 | nano-banana-2 | 1K / 2K / 4K |
| Nano Banana 2.1 | nano-banana-2.1 | 1K / 2K / 4K |
| Nano Banana Pro | nano-banana-pro | 1K / 2K / 4K |
- Supported ratios are 1:1, 16:9, 9:16, 4:3 and 3:4. Auto defaults to 1:1. The 4:5 ratio used in the tutorial's GPT example is not supported.
- This integration generates one image per request at quality=auto. Nano Banana reference-image editing is not currently supported.
- See Models & prices for current rates. Check the returned image dimensions and the logged charge. Actual dimensions and duration depend on the result.
Browser workspace and API
The workspace provides model, prompt and resolution options. Download generated images and manage them in the library. Developers can use the compatible image-generation endpoint.
- Text to image: start with a description, generate a version, then adjust details.
- Reference editing: choose a model that explicitly supports it and submit a reference with your instructions.
- API: use /v1/images/generations. Check the catalog for available models and supported parameters.
Check three things before generating
The same model name does not guarantee every parameter combination works. Check its current description and price first.
- Purpose: poster, cover, product image or character illustration.
- Capabilities: reference images, target size and chosen quality.
- Cost: the quote, your balance and billing rules. A larger size is not always the right choice.
Improve the result step by step
Start with the subject and composition, then adjust lighting and style, then local details. Keep each prompt and its parameters for comparison. Prompts are starting points, not promises of a fixed model output.
FAQ
Questions & answers
How do I choose an image or chat model?
Choose a model marked for image generation. Chat, image generation and reference editing are different capabilities; check each separately.
Can I generate a poster with Chinese text?
Text rendering depends on the model and prompt. You can generate the background and subject first, then lay out important text in a design tool for accuracy.
What if generation fails?
Read the workspace notice and usage log, and check the model, parameters and balance. Charges follow actual records. Provide a request or order number when asking for help, never an API Key.