Ordalin
Submit a tool

Reviewed tool profile

Nano Banana 2.1

Generate and edit images with multi-image references, text rendering and up to 4K output.

Sources checked

Visit website ↗
Pricing
The Gemini API is usage-priced in USD. Standard image output starts at $0.0336 for a 1K image; Batch image output starts at $0.0168. Input tokens and text or thinking output are charged separately. The pricing table lists no free API tier for this model. Official source 1 ↗
Available as
Available through Gemini API and Google AI Studio. Google also lists Gemini App, Search AI Mode, Google Ads, Flow and Stitch as distribution channels; the prices below cover Gemini API usage. Official source 1 ↗ · Official source 2 ↗
Primary group
Image & Design

What it does

Nano Banana 2.1 is Google’s image generation and conversational editing model, available as gemini-nano-banana-2.1 in the Gemini API. It supports 1K, 2K and 4K images, multi-image references, and Google Web and Image Search grounding. Google released it to general availability on October 6, 2026.

Official source 1 ↗ · Official source 2 ↗

Features

  • Generate and refine images. Create images and refine them through conversational editing with the gemini-nano-banana-2.1 model. Official source 1 ↗
  • Choose resolution and panoramic formats. Generate 1K, 2K or 4K images, with 1K as the default. Google reports fixes for tiling artifacts in 1:4, 4:1, 1:8 and 8:1 formats at 2K and 4K. Official source 1 ↗
  • Combine reference images. Use up to 14 reference images. Google specifies character consistency for up to four characters and object fidelity for up to ten objects. Official source 1 ↗
  • Create text-heavy visuals. Google reports improved text rendering and infographic layout accuracy. Posters and diagrams are among its stated use cases. Official source 1 ↗ · Official source 2 ↗
  • Ground images in search. Use Google Web and Image Search grounding to bring search context into image generation. Official source 1 ↗
  • Adjust thinking. Select minimal, medium or high thinking; medium is the default. Official source 1 ↗

Pricing & free limits

Gemini API — Standard

$0.0336 / $0.0504 / $0.0756USD per 1K / 2K / 4K output image

Image output is $30 per million tokens. Input costs $1.50 per million tokens; text and thinking output costs $7.50 per million tokens. These token charges are separate from the image output amounts. Official source 1 ↗

Gemini API — Batch

$0.0168 / $0.0252 / $0.0378USD per 1K / 2K / 4K output image

Image output is $15 per million tokens. Input costs $0.75 per million tokens; text and thinking output costs $3.75 per million tokens. The model supports Batch API. Official source 1 ↗ · Official source 2 ↗

Billing details

Google Web and Image Search grounding includes 5,000 search requests per month shared across Gemini 3.x models, then costs $14 per 1,000 requests. One model request can run multiple billable search queries. Official source 1 ↗

For paid Gemini API use, Google says prompts and responses are not used to improve its products. Prompts and responses may still be logged for a limited period for abuse prevention and required legal or regulatory disclosures. Official source 1 ↗

Free access & limits

Google lists the model’s Standard and Batch API free tiers as unavailable. The page’s Google AI Studio trial link does not establish a recurring free image allowance for this model. Official source 1 ↗ · Official source 2 ↗

Best for & limitations

Iterate on an image through several rounds of conversational edits. Official source 1 ↗ · Official source 2 ↗

Draft posters, diagrams and localized text visuals, then check the rendered wording and small text. Official source 1 ↗

Combine reference images for scenes involving recurring characters or products, while checking consistency in the result. Official source 1 ↗ · Official source 2 ↗

  • Small text and long copy can degrade. Google reports poor rendering of small text, long paragraphs and page-length text; small text is often blurry at 1K. Official source 1 ↗
  • References do not guarantee consistency. Characters may differ between reference images and the output. An edit can also retain the subject’s original pose. Official source 1 ↗
  • Localized edits may miss instructions. Mask- or doodle-based edits may only partly follow instructions or leave ink marks. Left/right spatial placement can also be confused. Official source 1 ↗
  • Check facts and allow for failed requests. Google lists hallucinations, limits in factuality and 3D reasoning, and occasional slowness or timeouts. Official source 1 ↗
Nano Banana 2.1 product screenshot
Website previewFixed desktop viewport · 1440 × 900

Similar tools

Published tools that share its categories.

PixfyCreate an AI character from text or a reference image and reuse it across outfits, poses, scenes, products, and brand mascots.
Image & DesignFree planChecked Oct 1
Visit
VirseCreate and edit commercial visual work on a canvas with multimodal AI models and aesthetic memory.
Image & DesignFree planChecked Aug 27
Visit