ai-toolstutorialsgeminihow-togoogle

How to Use Nano Banana 2.1 in the Gemini App

Step-by-step guide to generate and edit images with Nano Banana 2.1 in Gemini. Covers precise prompts, subject consistency, and practical tips.

October 20, 2026 Yum News

Google released Nano Banana 2.1 on 6 October 2026. The model improves visual quality, text rendering, mask-based editing, and subject consistency while keeping Flash-level speed. It runs in the Gemini app, AI Mode in Search, Google AI Studio, and the Gemini API.

If you already use the Gemini app for images, this update gives tighter control over what changes and what stays the same. For general Gemini app tasks beyond images, see our guide on how to use Gemini 4 Argon in the Gemini app.

What Nano Banana 2.1 Adds

Nano Banana 2.1 builds on Nano Banana 2 (Gemini 3.1 Flash Image) and sits on the Gemini 3.6 Flash architecture. Google lists these upgrades:

Better visual quality and realism at 1K, 2K, and 4K resolutions (default 1K).
Fixed tiling artifacts on wide and panoramic ratios such as 1:4, 4:1, 1:8, and 8:1.
Stronger text rendering and infographic layout accuracy.
Multi-image fusion with up to 14 reference images. Character consistency holds for up to four characters; object fidelity holds for up to ten objects.
Grounding via Google Web and Image Search.
Configurable thinking levels: minimal, medium (default), and high.

The model ID is gemini-nano-banana-2.1. Pricing for developers is roughly half the previous Nano Banana 2 rate for image outputs.

Person working on creative digital designs at a desk

Access Nano Banana 2.1 in the Gemini App

Open the Gemini app on Android or iOS, or go to gemini.google.com in a browser. Sign in with your Google account.

Look for the Create images option. It often appears as a tools chip or with the banana emoji. Select it to enter the image workspace.

Nano Banana 2.1 is rolling out as the default image model in the Gemini app. If you see a model selector, choose Nano Banana 2.1 or the latest Flash image option. Google AI Pro and Ultra subscribers can still regenerate with Nano Banana Pro via the three-dot menu on a result when they need higher fidelity for specialized tasks.

The feature is available to free users within normal Gemini limits. Exact daily image caps are not published as a fixed number; usage depends on your plan and current load.

Generate an Image from a Text Prompt

Start with a clear description that covers subject, setting, lighting, composition, and any required text.

Tap Create images or the banana chip.
Type your prompt. Example: “A square product photo of a matte sage-green ceramic mug on a pale oak desk. Soft morning window light from the left. The mug has a thick rounded rim and a large curved handle on the right. Leave clean space at the top for a short headline.”
Choose aspect ratio and resolution if the interface shows the options. Start at 1K while testing composition.
Generate the image.
Review the result. Use the chat or regenerate button to refine.

If the first output misses a detail, add the missing constraint in a follow-up message rather than rewriting the entire prompt.

Edit an Existing Image with Precise Control

Nano Banana 2.1 improves mask-based and conversational editing. Upload a photo or use a previously generated image, then describe the change and what must stay fixed.

Open the image workspace or upload a file via the plus button.
Write an edit instruction that separates the change from the keep list. Example: “Change only the background sky to a soft overcast grey. Keep the building, street, people, lighting on the facade, and all text exactly as they are.”
Generate the edit.
Zoom in to check edges, lighting match, and any text.
If the change spilled into protected areas, name a smaller region and retry.

You can chain edits. Take the new result and ask for the next single change. For multi-reference work, attach up to 14 images and label their roles in the prompt (for example, “character reference 1,” “product angle 2”).

Close-up of hands editing photos on a tablet

Practical Prompt Tips

Name one change at a time for edits. List the keep items explicitly.
Put required text in quotation marks and state where it appears. Example: “The only visible text is ‘CLEAR DAY’ on the label.”
For character consistency, supply two or three reference angles and end the prompt with “keep facial features, clothing colors, and proportions identical to the references.”
Use thinking level high when the scene is complex or when factual grounding (via Search) matters. Switch to minimal for quick drafts.
Start at 1K. Move to 2K or 4K only after the composition and text are correct.

Wide panoramic ratios now render cleanly at higher resolutions, so they are safer for banners and headers.

Watch a Walkthrough

Limits and Next Steps

Results can still show small errors in dense text, complex overlapping objects, or extreme lighting. Always review generated images before public use. Nano Banana 2.1 supports grounding with Google Search, which helps with real-world subjects, but it does not replace fact-checking.

Developers can call the model via the Gemini API in Google AI Studio. The same prompt patterns apply, with extra controls for resolution, aspect ratio, and thinking level.

Try a simple product edit or a short text poster today. Note what the model holds fixed and where it still needs a follow-up. That feedback loop produces better results than a single long prompt.