nano-banana-2-lite:official is a lightweight image creation model in the Google Nano Banana series, suitable for generating images from text and adjusting subjects, backgrounds, and styles using reference images. It is positioned for 1K image creation, focusing on concept drafts, asset variations, and interactive editing. For workflows that need to establish composition and visual direction before producing refined final work, Lite is a practical starting point.
Clarify capacity, inputs and outputs, and invocation methods before choosing a model.
Lite is positioned for 1K image creation; use aspect ratios, task processing, and result fields according to this platform's image interface.
Learn what nano-banana-2-lite:official can bring to your work.
Organize prompts using descriptions of the subject, environment, lighting, composition, and style, and Lite can be used to explore different visual directions for the same idea. It is suitable for validating visual relationships early in the design process, such as comparing centered versus side-positioned products and bright versus dark backgrounds, rather than placing the burden of a refined final result on the task from the start.
When editing, provide the original image and distinguish in the prompt between the parts to retain and the parts to change, for example, retaining a cup's shape and color while replacing only the tabletop and background. Combining multiple images allows assets such as subjects and scenes to participate in creation together; clearly explaining the purpose of each reference image helps reduce element mixing.
Seven aspect ratios cover square assets, landscape covers, and portrait displays. When choosing a ratio, describe both the subject position and the direction of negative space, for example, placing a person lower in a vertical image while leaving room for a title at the top. This designs the composition during generation rather than mechanically cropping the same image into different layouts.
Start with specific tasks to find where the model can be effective.
Input product reference images, describe scenes such as a wooden table, a window side, or a minimalist interior, and require the product outline and primary color to be preserved. The deliverables are scene candidate images for discussion, making it easier for teams to compare backgrounds, lighting, and placement. When details such as packaging text or port locations are involved, they should be checked item by item against the physical product.
Input the event theme, brand colors, and target aspect ratio, first generate cover images or poster backgrounds with different compositions, then adjust the atmosphere and elements around the selected direction. Lite is better suited to producing visual concept candidates; formal titles, dates, and rules text can be added during the layout stage to avoid making image generation responsible for all text proofreading.
Use the same character image as a reference, separately describe changes to the background, clothing colors, or visual style, and create drafts for stickers, mood images, and content illustrations. Clearly specify that facial features and key identifiers should be retained in every round, and check whether the results deviate from the original design; reference images can help maintain continuity of appearance, but they do not lock in every detail.
Choose based on task complexity, input materials, and expected results.
If the task focuses on finding suitable composition, color schemes, and scenes, and 1K images are sufficient for preview and discussion, Lite can be prioritized. It is in a different tier from nano-banana-2: Lite emphasizes lightweight creation and rapid validation, while nano-banana-2 is geared toward a balanced choice. Decide based on delivery requirements rather than treating the two names as the same model.
If 2K or 4K delivery, complex detail presentation, or more refined brand visuals are needed, consider nano-banana-pro. Do not substitute an upgrade by setting a higher resolution value for Lite. A more practical workflow is to use Lite to validate the concept first, then provide the selected composition requirements and reference materials to a model suitable for final production.
From a small-scale task to formal integration.
Define the goal, required inputs, and output requirements, using real business examples as a starting point.
Open the trial page, confirm the parameters supported by this entry point, then submit a small-scale task to review the results.
Keep the complete model ID, use the request format specified in the documentation, and confirm billing rules on the Pricing page.
Before formal use, understand the output quality and capability scope.
Answers to common questions about using nano-banana-2-lite:official.
It is suitable for 1K visual drafts, product scene-change trials, campaign asset exploration, and reference image editing. It is especially useful for workflows that compare multiple creative directions before deciding on a final solution. If the task primarily evaluates print dimensions or extremely fine textures, consider a higher-resolution model first.
Lite is intended for 1K image creation and should not be treated as having 2K or 4K output options. When higher resolution is needed, choose a model that supports the required size, such as nano-banana-pro; changing only the resolution value in the request cannot substitute for model capability.
Use the edit action, place the reference image in image_urls, and clearly state in the prompt which subject characteristics must be preserved and what needs to change in the background. Avoid changing too many attributes at once; complete the scene change first, then check whether the subject's outline, colors, and details have changed in unwanted ways.
You should also adjust the composition description accordingly. For landscape images, emphasize left and right negative space and the expanded environment; for portrait images, emphasize the subject's vertical placement and space at the top. Lite provides seven aspect ratios, so choose a frame that fits the display placement and then describe the element layout, which is more targeted than cropping after generation.
Send a POST request to /nano-banana/images, specifying the model, action, and prompt; add image_urls when editing. The response provides a task identifier and image_url in data, which can be used for display or download. For asynchronous processing, you can use async and callback_url to arrange result delivery.
Model information · Updated: 2026-10-01. See the API and pricing sections for request parameters and billing rules.
Start with a clear goal and judge from actual results whether it fits your work.
Pricing
$0.0172· per generation
You pay$13,584for 142,800 credits · $0.0951 / credits46% OFF
You're billed in credits. Pick a top-up tier to see its per-credit rate — the larger the top-up, the cheaper each credit.
This is only a basic call example. See the full docs for more parameters and advanced usage.
View Docs