Google has released Nano Banana 2.1, the latest version of its image generation and editing model. It became generally available in the Gemini API on 6 October 2026 under the model ID gemini-nano-banana-2.1, and it arrived with a headline number: in the API, each 1K and 2K image now costs half of what it did with Nano Banana 2. Below we cover what changed, where you can use it, what it really costs (including an important caveat about 4K pricing) and how to write prompts that get good results.
What is Nano Banana 2.1 and what changed from Nano Banana 2?
Nano Banana is the popular name for Gemini's image model family. Nano Banana 2 (officially Gemini 3.1 Flash Image) launched in February 2026, while Nano Banana Pro (Gemini 3 Pro Image) remains the top-quality option. Version 2.1 is an update to the Flash model: Google describes it as its high-efficiency model for image generation and conversational editing, keeping Flash-tier speed and cost while lifting quality.
According to Google's developer documentation, the main improvements are:
- Better visual quality and prompt adherence, especially for framing and composition.
- Legible in-image text for posters, infographics and labels, historically a weak spot for image models.
- Multi-turn character consistency: when you edit over several steps, faces and outfits stay the same.
- Up to 14 reference images (up to 4 characters and 10 objects) that can be fused into a single scene.
- Configurable thinking levels (minimal, medium by default, high) and Google Search grounding on web and image results for more accurate real-world subjects.
- 1K, 2K and 4K output, with a fix for the tiling artifacts seen in very wide or tall formats (1:4, 4:1, 1:8 and 8:1).
It accepts text, images, video and PDF as input. As with the rest of the family, outputs carry Google's invisible SynthID watermark.
Where to use Nano Banana 2.1: Gemini app, AI Studio and the API
Google is rolling it out across the Gemini app, AI Mode in Search, Google AI Studio, the Gemini API, Flow, Stitch, Google Ads and its Google Cloud enterprise platform. The free versus paid split matters, though:
- Gemini app, free tier: still runs Nano Banana 2, free with daily limits. Google's Gemini page says Nano Banana 2.1 is unlocked with a Google AI plan.
- Gemini app, Google AI Pro, Plus and Ultra: Nano Banana 2.1 by default, higher generation limits and a "Redo with Pro" option that regenerates an image with Nano Banana Pro.
- Google AI Studio and the API: available to developers, but the official pricing table lists no free tier for this model, so every image is billed.
Bottom line: if you want Nano Banana for free, the practical route is still the Gemini app with Nano Banana 2, which is already very capable. If you specifically need 2.1 at volume, paying per image through the API is usually cheaper than a monthly subscription.
Nano Banana 2.1 pricing: how much does each image cost?
These are the per-image prices Google lists on its official Gemini API pricing page (USD, as of 14 October 2026):
| Resolution | Nano Banana 2 | Nano Banana 2.1 | 2.1 Batch |
|---|---|---|---|
| 1K | $0.067 | $0.0336 | $0.0168 |
| 2K | $0.101 | $0.0504 | $0.0252 |
| 4K | $0.151 | $0.113 | $0.0567 |
Two details that a lot of coverage missed:
- 4K is not half price. A figure of $0.0756 per 4K image circulated widely, but Google's official table shows $0.113, a cut of roughly 25%, not 50%. The 50% saving does hold at 1K and 2K.
- Input got more expensive. Input tokens (text, image or video) rise from $0.50 to $1.50 per million, and text and thinking output from $3 to $7.50 per million. If you send many reference images or use high thinking, your real saving will be smaller than the table suggests.
For context, Nano Banana Pro costs $0.134 per image at 1K or 2K and $0.24 at 4K, so one Pro image at 1K buys you almost four 2.1 images at the same resolution. Batch mode (deferred processing, no instant response) halves the price again and is the obvious choice for product catalogues or bulk variations.
Google has also marked gemini-3.1-flash-image as deprecated in the API and recommends migrating to 2.1, though no shutdown date has been announced yet.
How to use Nano Banana 2.1 step by step
In the Gemini app
- Go to gemini.google.com or open the mobile app.
- Pick the image creation tool or simply ask "create an image of...".
- To edit, upload a photo and describe the change: "replace the background with a bright office" or "put this logo on a mug".
- Keep iterating in the same chat; the model remembers the previous image.
In Google AI Studio or the API
- Open aistudio.google.com with your Google account and select gemini-nano-banana-2.1.
- Set resolution (1K, 2K or 4K), aspect ratio and thinking level.
- Once your prompt works, create an API key with billing enabled and wire it into your app. For large volumes, use the Batch API.
Nano Banana prompt tips that actually help
Google's own Gemini guide recommends starting with a simple formula and then adding detail. In practice, these habits pay off:
- Describe a scene, not a keyword list: subject, action, setting, lighting and style in full sentences. "Portrait of a baker smiling in her bakery at dawn, warm side light, 50mm photo" beats "baker, photo, light".
- Put text in quotes and say where it goes: headline "AUTUMN SALE" at the top in white sans-serif type. This is where 2.1's text improvements show.
- State the format up front: 16:9 for YouTube thumbnails, 9:16 for stories, 1:1 for social posts.
- Refer to inputs by number: "the woman in image 1 holds the bag from image 2". With several references, numbering avoids mix-ups.
- Edit in small steps: one change per message keeps consistency better than five changes at once.
- Use high thinking for infographics, diagrams and busy scenes, and minimal for quick sketches.
For more on writing good instructions, see our list of the best AI prompts for productivity. If you want to animate your images afterwards, read about Wan 2.6, which turns photos into video. And to choose a general assistant, check our Claude vs ChatGPT vs Gemini comparison.
Official sources: Nano Banana 2.1 model documentation, Gemini API pricing and Gemini image generation page.
Frequently asked questions
Is Nano Banana 2.1 free?
Not entirely. Free Gemini app users get Nano Banana 2, while 2.1 requires a Google AI Pro, Plus or Ultra plan. In the API you pay per image, starting at $0.0336 for 1K.
What is the difference between Nano Banana 2.1 and Nano Banana Pro?
Nano Banana 2.1 is the fast, low-cost Flash model; Pro is the top-quality model and costs about four times more per 1K image. For most everyday work, 2.1 is enough.
How much does a 4K image cost with Nano Banana 2.1?
Google's official pricing lists $0.113 per image in standard mode and $0.0567 in Batch mode. The widely shared $0.0756 figure does not match the pricing page.
Do Nano Banana images have a watermark?
Yes. Google embeds SynthID, an invisible watermark that identifies the image as AI-generated.
