Nano Banana (Gemini 2.5 Flash Image)
Nano Banana is Google's Gemini image generation and editing model, known for consistent characters and natural-language edits.
Try Nano Banana online → Full guide on GitHub
Overview
Nano Banana is the nickname, and now the official product name, for Google's Gemini image generation and editing models. It began as Gemini 2.5 Flash Image in August 2025, was followed by Nano Banana Pro (Gemini 3 Pro Image) in November 2025 and Nano Banana 2 in early 2026. The model is known for character consistency across edits, mask-free local editing from natural-language prompts, multi-image fusion and, on the Pro tier, 4K output with reliable text rendering. It is available in the Gemini app, Gemini API and Vertex AI.
What it can do
- Prompt-based local edits without masks
- Character and product consistency across edits
- Multi-image fusion, up to 14 references on Pro
- 1K, 2K and 4K output on Nano Banana Pro
- Reliable in-image text rendering on Pro
- Multi-turn conversational refinement
Versions
| Version | Released | Notes |
|---|---|---|
| Gemini 2.5 Flash Image (Nano Banana) | 2025-08 | First release; appeared anonymously on LMArena as nano-banana, then shipped as gemini-2.5-flash-image-preview. Native 1024 px output, up to 3 input images, SynthID watermark. |
| Gemini 2.5 Flash Image GA | 2025-10 | Stable model ID gemini-2.5-flash-image; adds the ten selectable aspect ratios and general availability on Vertex AI. |
| Nano Banana Pro (Gemini 3 Pro Image) | 2025-11 | Built on Gemini 3 Pro with a thinking step; 1K/2K/4K output, up to 14 reference images, Google Search grounding, much better text rendering. |
| Nano Banana 2 (Gemini 3.1 Flash Image) | 2026-02 | Flash-tier successor that brings much of the Pro quality and higher resolution to the faster, cheaper model. |
How to try Nano Banana (Gemini 2.5 Flash Image)
Nano Banana is available in the Gemini app (free with a daily limit, higher quotas on Google AI Pro and Ultra), through the Gemini API in Google AI Studio using the model IDs gemini-2.5-flash-image and gemini-3-pro-image-preview, on Vertex AI for enterprise, and inside Google Search, Workspace, Adobe Firefly and Photoshop. Availability depends on country and plan. The quickest way to try it without a Google account or subscription is the pay-per-generation service at https://banana-nano.co.
FAQ
What is Nano Banana?
Nano Banana is the nickname and now official product name for Google's Gemini image generation and editing models, starting with Gemini 2.5 Flash Image in August 2025. It is known for consistent characters, mask-free editing and multi-image fusion.
Is Nano Banana free?
Partly. The Flash tier is free in the Gemini app with a daily generation limit, and Google AI Studio has a limited free tier for the API. Nano Banana Pro and higher quotas require a paid Google AI plan or paid API usage.
Is there a Nano Banana API?
Yes. It is the Gemini API model gemini-2.5-flash-image, with gemini-3-pro-image-preview for Nano Banana Pro, available through Google AI Studio and Vertex AI, and resold by hosts such as fal.ai, Replicate and OpenRouter.
Does Nano Banana have an official GitHub repository?
No. Nano Banana is a closed, hosted model and Google has not released weights or an official repository. This page collects publicly available information about it.
How do I try Nano Banana online?
Open the Gemini app and attach or describe an image, or use Google AI Studio for the API. For pay-per-generation access with no account or waitlist, use https://banana-nano.co.
What are the limits of Nano Banana?
The Flash tier outputs at about 1024 px and is weak on small text; the Pro tier goes to 4K but is slower and costs more. All images carry a SynthID watermark, safety filters block some subjects, and the model does not generate video.
What is the difference between Nano Banana and Nano Banana Pro?
Nano Banana (Gemini 2.5 Flash Image) is the fast, low-cost tier with up to 3 input images and 1K output. Nano Banana Pro (Gemini 3 Pro Image) adds a reasoning step, 2K and 4K output, up to 14 reference images, Google Search grounding and far better text rendering.