Up to 14 reference images per generation, characters that stay consistent across scenes, and text that renders accurately inside your images. Fast enough for real production workflows — text-to-image or direct editing.
Please enter a description
Select a mode, upload an image, or enter a description on the left, and let AI create amazing visual works for you
Built for speed and iteration. Supports up to 14 reference images per generation and real-time Google Search grounding for content that stays current.
Better adherence to complex prompts means fewer retries. Describe exactly what you want, including multi-part conditions and specific style requirements, and the model delivers without cutting corners.
Upload up to 14 images per generation. Mix product shots, style boards, character designs, and layout references in one request. The more context you give, the closer the output lands.
Keep characters consistent across scenes. Face, outfit, and visual style stay locked no matter how the setting changes. Built for storyboards, comics, and recurring brand characters.
When it matters, the model pulls current information from Google Search before generating. Today's date, live events, trending products. Output grounded in what's actually happening.
Precise text inside images. Headlines, labels, and product copy rendered with sharp letterforms and accurate spelling. A consistent weak point in AI image models that Nano Banana 2 handles reliably.
Sharp details, accurate lighting, rich textures. Visual quality on par with Nano Banana Pro, with the speed expected from a Flash-tier model.
A look at what Nano Banana 2 can do across different use cases.
Stop getting a different-looking character every time. Nano Banana 2 keeps the same face, outfit, and art style locked across scenes, angles, and actions. Built for serialized content, IP development, and storyboard production.
Most image models stop at their training cutoff. With Google Search grounding, Nano Banana 2 can accurately depict newly released products, current events, and niche subjects — including things that barely existed at training time.
Generate posters, magazine covers, and brand assets with correctly spelled, readable text already baked in. Nano Banana 2 renders headlines, slogans, and labels sharp and accurate from the first generation — no post-editing needed.
Four steps from idea to finished image.
Choose text-to-image or image editing mode. Select model, aspect ratio, and resolution.
Write your prompt, upload up to 14 reference images, or combine both.
Hit generate and review the result. Type a follow-up to adjust without starting over.
Save the result, or type a follow-up to keep refining until it's right.
Common questions about using Nano Banana 2.
Nano Banana 2 is Google's image generation model built on Gemini 3.1 Flash Image. Because it inherits Gemini's broad world knowledge, it handles real-world concepts, current events, and detailed creative briefs better than most image models. Generation is fast without sacrificing detail, and it adds native support for multi-image fusion (up to 14 references), follow-up edits through conversation, and Google Search grounding.
Nano Banana Pro is optimized for maximum quality — fine detail, photorealistic rendering, and high fidelity outputs. Nano Banana 2 trades some of that ceiling for noticeably faster generation and lower cost per image, while adding multi-image input and conversational editing. For most production workflows where you're iterating frequently, Nano Banana 2 is the more practical choice.
Up to 14 images per generation. You can combine product photos, style references, character designs, and layout examples in a single request. Nano Banana 2 figures out how to use each one from your prompt, so you don't need to label or explain each image individually.
Built on Gemini 3.1 Flash Image, Nano Banana 2 has broad knowledge of the real world baked in. It understands current brands, recognizable locations, cultural references, and factual contexts without needing extra explanation. For content that requires up-to-date information, Google Search grounding can pull in live data before generating.
With search grounding enabled, Nano Banana 2 queries Google Search before generating to incorporate current information — today's date, recent events, trending products, and more. This is useful when your image content needs to reflect what's happening now rather than what the model learned during training.
Credits are charged by output resolution: 1K costs 8 credits, 2K costs 12 credits, and 4K costs 18 credits. New accounts start with 20 free credits, which is enough to try Nano Banana 2 at different resolutions and see what works for your use case.
Nano Banana 2 supports 1K, 2K, and 4K output resolutions. For aspect ratios, it covers the full range from square and portrait to landscape and ultra-wide formats, including extreme ratios like 1:8 and 8:1 that most models don't support. This makes it well suited for everything from social media posts to wide banners and vertical story formats.
Yes. Headlines, labels, product copy, and captions render with accurate letterforms and correct spelling. Precise in-image text is a weak point for most AI image generators, and Nano Banana 2 handles it reliably. It also supports translating existing text within an image into another language, which saves time when producing region-specific versions of the same asset.
Yes. Images generated through this platform can be used for commercial purposes including product design, advertising, client projects, and resale. Check our terms of service for the complete details.
Sign up for an account and you get 20 credits instantly, no credit card required. Use them to try different resolutions, upload reference images, and test follow-up edits. Once you have a feel for how it works, you can decide whether to add more credits.
Run Nano Banana 2 directly in your browser. Iterate fast, keep characters consistent across scenes, and get readable text inside your images — no extra tools, no setup. 20 free credits to start, no card needed.
