How to Use Nano Banana (Gemini Image Generation) in the Gemini App, Flow and the Gemini API (October 2026)

Quick answer: Nano Banana is Google’s name for Gemini’s built-in image generation and editing. You can use it three ways: type a prompt in the Gemini app (image generation and editing is on the free plan), use it inside Google Flow, or call it from the Gemini API, where Google recommends Nano Banana 2.1 (gemini-nano-banana-2.1) for new projects. Every image carries an invisible SynthID watermark.

Which Nano Banana model should you use?

Google’s image generation guide (accessed October 25, 2026) lists these models:

Model	Model ID	Best for
Nano Banana 2.1	gemini-nano-banana-2.1	The default choice: fast, better text rendering, multi-turn editing, Google Search grounding, 1K/2K/4K output
Nano Banana 2 Lite	gemini-3.1-flash-lite-image	Google calls it the fastest and cheapest; not optimized for many reference images or multi-turn edits
Nano Banana 2	gemini-3.1-flash-image	Previous generation of the Flash image model
Nano Banana Pro	gemini-3-pro-image	The premium choice for the most complex visual work, brand consistency and localization
Nano Banana	gemini-2.5-flash-image	The legacy original; Google’s suggested migration path is Nano Banana 2 Lite

Per-image prices differ by model and resolution; see the Gemini API pricing page.

How do you use Nano Banana in the Gemini app?

1. Open gemini.google.com or the Gemini mobile app and sign in with a Google account.
2. Describe the image you want, for example “a flat-lay photo of a breakfast table, soft morning light, 4:5”.
3. To edit, upload a photo and say what to change: “replace the background with a beach at sunset, keep the person unchanged”.
4. Keep refining in the same chat. Nano Banana keeps the context of earlier turns.

Google’s subscriptions page lists image generation and editing on the free plan. Google AI Plus, Pro and Ultra give more access. See Google AI Pro vs Ultra for the plan differences.

How do you use Nano Banana in Google Flow?

Google Flow is Google’s AI creative studio. Its prompt box defaults to Nano Banana Pro for images, and the same project can then animate stills with Veo. Free accounts get limited Nano Banana Pro access in Flow, and paid plans add monthly Flow credits. The steps for switching Flow between image and video models are in our Veo 3.1 guide.

How do you call Nano Banana from the Gemini API?

Create a key in Google AI Studio (step by step: how to get a Gemini API key), set it as GEMINI_API_KEY, install the SDK (pip install google-genai), then:

from google import genai
import base64

client = genai.Client()
interaction = client.interactions.create(
    model="gemini-nano-banana-2.1",
    input="A minimalist poster of a banana-shaped rocket, bold title text 'NANO'",
)
with open("poster.png", "wb") as f:
    f.write(base64.b64decode(interaction.output_image.data))

To edit the result, send a follow-up with previous_interaction_id=interaction.id. Set response_format={"type": "image", "aspect_ratio": "16:9", "image_size": "2K"} to control the shape and resolution.

What can Nano Banana do?

From Google’s documentation:

• Text-to-image and image editing, including multi-turn conversational edits.
• Up to 14 reference images mixed into one result, for consistent characters or products.
• Aspect ratios including 1:1, 3:2, 2:3, 3:4, 4:3, 4:5, 5:4, 9:16, 16:9 and 21:9.
• Legible text for posters, menus, infographics and diagrams.
• Grounding with Google Search, to base images on current data such as weather or recent events (not on Nano Banana 2 Lite).
• Thinking mode, which drafts interim “thought images” before the final output.

Limits to know: there is no audio input, every image carries SynthID, and Google’s Prohibited Use Policy applies. Ready to animate a still? See image to video AI.

How do you write a good Nano Banana prompt?

Describe the scene rather than listing keywords. Name the subject, setting, lighting, camera angle and style, and give the aspect ratio. For edits, say what must stay unchanged (“keep the face and clothing identical”). For text in images, put the exact words in quotes and say where they go.