Add a prompt or reference
Describe the result clearly. For an edit, upload the image you have the right to process and state what should change and what should remain.
Google image model guide
Nano Image uses Google’s gemini-3-pro-image model—officially described as Nano Banana Pro—for its general image generation and image-to-image workflow. Upload a reference, describe the intended change, select an available output size, and review the generated result before download.
Sign-in and sufficient credits are required to process an image. Promotional credits depend on the active offer. Current pricing and output rights are shown before generation and on the pricing page.

Deployed model
gemini-3-pro-image
Official name
Nano Banana Pro
Inputs
Prompt, reference image, or both
Output choices
1K trial; paid 2K and 4K access
“Nano Banana” is a model nickname, not the Nano Image brand. The current server configuration uses gemini-3-pro-image, which Google identifies as Nano Banana Pro in its official model documentation. This page describes the real model connected to the product rather than using “Nano Banana” as a generic label for every image tool.
The AI art image generator accepts a written request, an optional reference image, and an available output resolution. With a reference image, the request can preserve the subject while changing style, mood, setting, lighting, or other described elements. Without a reference, it can create a new image from text. The output is generative, so it should be reviewed for accuracy, rights, safety, and fitness for the intended use.
This model is separate from the Qwen endpoint used for the dedicated camera-angle generator. Use Nano Banana Pro for general image creation and editing. Use Qwen camera controls when the specific task is to request a front, side, back, elevated, low, wide, or close-up alternate view from one source image.
Practical workflow
Describe the result clearly. For an edit, upload the image you have the right to process and state what should change and what should remain.
Trial access uses the Quick path. Purchased credits or a subscription unlock paid output rights and the available Standard or Best sizes.
Check subject identity, composition, text, hands, small details, and brand elements. Download only a result that meets your requirements.
A useful image-editing instruction names the subject, the requested change, and the elements that should stay stable. For example: “Keep the shoe shape, sole pattern, and logo placement. Replace the background with a warm studio gradient and add a soft shadow below the product.” That is easier to verify than an instruction such as “make it better, premium, cinematic, amazing.”
For creative transformations, state the medium or visual treatment, lighting direction, palette, and composition. For product or business work, protect measurable details in the instruction and compare them after generation. AI can alter label text, small logos, counts, packaging geometry, or fine material texture. If exact typography is required, it is usually safer to add final text in a design tool after the image is approved.
Nano Image separates the low-cost trial experience from paid output rights. Free and promotional credits use the Quick 1K path with a trial watermark. A subscription or purchased credit pack unlocks the paid output path, watermark-free eligible output, commercial-use rights for eligible results, and the ability to select 2K or 4K. The current application prices Standard 2K at 3 credits and Best 4K at 6 credits; the editor shows the selected cost before submission.
Promotional signup credits add trial capacity but do not by themselves unlock paid resolution or commercial rights. This prevents a temporary campaign from silently changing the value of purchased access. Because pricing and promotions can change, use the pricing page and the live selector as the source of truth at the moment you generate.
General image editing is only one part of the site. Use image cleanup when you want to brush over a localized object or distraction. Use the background remover when the requirement is a transparent cutout. Use the AI product photo editor when the workflow continues into white backgrounds, resizing, padding, shadows, or marketplace preparation.
These focused tools reduce ambiguity. A broad generative prompt may redraw more than the user intended, while a mask-based cleanup or background cutout has a narrower goal. Nano Image is an independent product and is not affiliated with or endorsed by Google. Google, Gemini, and Nano Banana Pro are used here to identify the model technology behind the feature.
Important limitations
Upload first, choose the controls that match your task, then review the generated view before using it in production.