Model · Updated October 13, 2026

Nano Banana Pro (Gemini) for AI agents

Nano Banana Pro is Google's Gemini 3 Pro Image model. AnyCap exposes text-to-image and image-to-image generation through its HTTP API and CLI. For a revision, supply your existing image and an edit instruction, save the returned output, then compare both files before using the result.

Generated example

Before-and-after proof for the revision workflow

This published example starts from a Nano Banana 2 draft and uses Nano Banana Pro to revise its background, composition and lighting. Compare the device itself as well as the surrounding scene.

Rough tabletop product shot of a wearable device on a cluttered desk with mixed lighting. — Before

Refined product marketing shot of the same wearable device centered on a minimal pedestal with a warm cream studio background. — After

The left image is a rough Nano Banana 2 draft. The right image is the Nano Banana Pro revision generated through AnyCap from that source asset.

Edit prompt used with AnyCap

replace the cluttered desk with a warm cream studio gradient and a minimal pedestal, center the wearable device, add soft rim light and cleaner composition, preserve the device shape, make it look launch-ready, premium product marketing photo, no text, no watermark

When agents should choose Nano Banana Pro

Revise an existing image with a text instruction
Adjust lighting, composition, or background without rebuilding the asset
Keep the original and revised files together for review
Start from an input image instead of a blank prompt

Call Nano Banana Pro through AnyCap

Check the current editing schema

anycap image models nano-banana-pro schema --operation generate --mode image-to-image

Edit an existing image

anycap image generate --model nano-banana-pro --mode image-to-image --prompt "Replace the background with warm studio light; preserve the subject" --param images=./input.png -o revised.png

Retry an existing output download

anycap download "<returned-output-url>" -o revised.png

Workflow tradeoffs

Choose Nano Banana Pro
For the reference-image workflow shown here, keep the input file and ask for a specific change. Review the output against that reference.
Compare with Seedream 5
Both models expose text-to-image and image-to-image through AnyCap. Compare the same brief or reference, current request limits and actual charges before choosing.

Nano Banana Pro vs nearby choices

Dimension	Nano Banana Pro	Alternative
Input modes	Text-to-image and image-to-image	Seedream 5 also exposes both modes
Reference-image limit	Up to nine images in the checked AnyCap editing schema	Seedream 5's checked editing schema also allows up to nine images
Resolution settings	1k, 2k and 4k in the checked AnyCap schema	Seedream 5 exposes 2k and 4k; inspect each model's current schema

FAQ

What is Nano Banana Pro best for?

This guide uses it for reference-image editing. AnyCap also exposes text-to-image. Choose using the task, current schema, observed output and cost.

How do agents call Nano Banana Pro through AnyCap?

Use the AnyCap skill and CLI with your reference image and an edit instruction, or connect through the AnyCap API. Follow the image documentation for setup and supported inputs, then save and review the returned file.

Should I use Nano Banana Pro or Seedream 5?

Both support generation and editing in AnyCap. Compare a single matching request, current resolution and reference limits, and the actual returned charge. The example pair does not establish a universal model ranking.

Try one image edit Access and delivery Image generation Seedream 5

From a reference image to a reviewed edit

Keep the original
Choose the reference image you want to revise. Save an untouched copy so you can compare the product details after editing.
Request one specific change
Describe the background, lighting or composition change, and name the details that must remain. Confirm current pricing before generating a revision.
Save and compare both versions
Open the returned image and compare it with the original. Check the subject outline, framing and unwanted text, then review the actual charge before requesting another variant.

If an output was returned but saving failed, download that same output again. If submission timed out and its result is unknown, check the recorded request before submitting another billed generation.

For API access, create an AnyCap key. Check pricing before generating, and follow the image documentation for setup, reference inputs and file delivery.

Check cost, inputs and the delivered file

The production catalog and editing schema were checked on October 12, 2026. They expose an active nano-banana-pro model, up to nine reference images, 1k/2k/4k resolutions and PNG/JPG/WebP formats. A schema response is not a price quote.

The before-and-after pair is an existing published example. It illustrates one revision without establishing a current speed or cost benchmark. Keep both files and review the result before using it.

Edit one reference image with Nano Banana Pro

A revised.png file from your own ./input.png reference. Compare it with the original before requesting more variants.

Install the CLI and sign in once. If you are already configured, continue with the access and cost checks.

curl -fsSL https://anycap.ai/install.sh | sh
anycap login
anycap --version
anycap status

Run anycap status, inspect the current model schema, and check pricing and your balance. Generation is billed; model access and rates can change.

anycap image models nano-banana-pro schema --operation generate --mode image-to-image

Place your reference at ./input.png or replace that path with your own file.

Run this command after checking the schema and cost. Inspect the delivered file and actual charge.

anycap image generate --model nano-banana-pro --mode image-to-image --prompt "Replace the background with warm studio light; preserve the subject" --param images=./input.png -o revised.png

Success means the output file exists and can be opened. Compare it with your brief or reference, including dimensions, subject details and unwanted text. If generation returned an output URL but saving failed, retry downloading that output before requesting another image.

Read the related workflow. For setup help, see installation and login.

CLI version and updating · 0.6.5

Current release checked October 21, 2026: 0.6.5. Use the update command for your installation method, then verify the version and login in the same environment that runs your agent.

Direct binary installation

anycap update

npm installation

npm install -g @anycap/cli@latest

anycap --version
anycap status

Version 0.6.5 retries transient downloads, recovers an expired video link by querying the same task, and saves multiple image outputs to numbered files. After a video download failure, resume the existing task instead of generating again.

anycap video tasks wait <task-id> -o result.mp4

Dated examples retain the CLI version, output and charge from the original run. Updating the CLI does not revalidate an old result. Check the live model schema and current pricing before a new billed task.