Google Image generation and reference-led iteration
Nano Banana 2
Take one product concept from a studio draft to a set of reviewable campaign images.
Google calls Nano Banana 2 Gemini 3.1 Flash Image. On TokenHot, use nano-banana-2 to generate campaign candidates from a brief or use a reference photo in the Playground.
- Image generation
- Reference-led iteration
- Format variants
This route is Nano Banana 2. Google recommends the newer Nano Banana 2.1 for new Gemini projects. Use the TokenHot example to confirm the request shape and review each image for subject drift, misplaced details and legibility.
Try this model
Choose an aspect ratio and resolution to generate an image.
nano-banana-2Input
Generated result
Enter a prompt, choose your settings, and generate.
USD /image
Pricing
| Group | Price |
|---|---|
| default | $0.03 |
Public list prices are shown in USD. Final charges may vary by account group and usage tier.
Overview
What is Nano Banana 2?
Nano Banana 2 is useful for building a small asset family: settle the product angle and lighting in one candidate, then adapt the brief for a studio shot, a lifestyle image and a portrait crop. Review every variant next to the same source photo to catch changes in the silhouette, materials and small product details.
The TokenHot API example shows a text prompt with a 16:9 aspect ratio and 2K image size. For a reference-led draft, the Playground lets you attach one image. Confirm the image payload and any required settings before automating edits through the API.
Anchor the set to one source
Keep the same product reference and a short list of features that must remain stable across the set.
Change one part of the brief
Ask for a background change while keeping the subject, angle and proportions fixed, then compare against the source.
Design for each placement
Recompose for each channel so the product and space for a headline survive at the final aspect ratio.
From brief to deliverable
Put Nano Banana 2 to work
Ecommerce and product marketing teams
Build a product image set around one reference
Use an approved product photo to guide a studio packshot, a lifestyle composition and a tighter crop. Judge each against the same source before choosing a set.
- Tools you need
- A rights-cleared source photo, image generation and side-by-side review.
- Keep in mind
- Try one reference photo per request in the Playground. Reject variants that change the product geometry or packaging.
View a task brief
Create one candidate for this placement using the supplied product reference. Keep the product geometry, trim and material unchanged. Follow the requested crop and use the specified lighting and environment. Leave the product markings exactly as shown or ask for a cleaner reference.
Creative production and ecommerce teams
Test a background replacement without redesigning the product
Attach the original frame in the Playground and describe the new surface and shadow. Compare the candidate with the source at the same scale.
- Tools you need
- An image reference workflow and a visual comparison tool.
- Keep in mind
- An edit instruction guides the model but does not lock pixels. Retouch or reject a result if straps, seams, or product colors drift.
View a task brief
Use this frame as the reference. Replace the white surface with light gray stone and a soft contact shadow. Preserve the product silhouette, angle, straps, zipper and material. Keep the product centered at the same size.
Brand and design teams
Compose for wide, square and vertical placements
Write a separate composition brief for each channel. Keep the subject and lighting consistent while making space for copy in each layout.
- Tools you need
- A format specification, layout editor and final crop review.
- Keep in mind
- Designing for a target ratio takes a separate composition check. Add final copy in the layout editor where exact text and margins can be inspected.
View a task brief
Compose an image for the supplied placement dimensions. Keep the approved product styling and light direction. Place the product so the designated copy area remains clear. Leave the image free of text and check the composition at the final crop.
Capabilities in context
Key features
| Feature | What you get | Why it matters |
|---|---|---|
| Native request format | TokenHot Gemini generateContent route | Use nano-banana-2 in the endpoint path and request an image response in generationConfig. |
| Framing and size | The API example uses 16:9 and 2K | Set them for the intended placement and confirm the accepted values before building a batch workflow. |
| Reference workflows | One reference image in the TokenHot Playground | Use the reference to guide one draft. Confirm the API payload before automating edits. |
| Version identity | Nano Banana 2 and Nano Banana 2.1 are separate model IDs | Pin the exact model in requests and compare on a shared set of references before migrating. |
The three scenario illustrations were created separately with ImageGen. Use your own inputs to evaluate actual Nano Banana 2 output. TokenHot Nano Banana 2 API · Google image generation guide
Audience & fit
Where Nano Banana 2 fits
Ecommerce teams
Produce candidate product images for different listing placements while checking consistency against the reference.
Brand designers
Explore a shared visual direction across formats and add final copy during layout.
Creative developers
Compare route parameters and rejected assets before adding image jobs to a production flow.
Cost context
Plan the cost of a useful result
Estimate a complete set
Count the variants and revision calls needed for each usable set. Read the current route pricing before launching batches.
Pick the delivery size
Use the lowest supported size that fits the placement and reserve larger outputs for assets that benefit from the detail.
Review before scaling
Check one product across all intended placements, then record failed calls and rejected images when budgeting the rest.
Use the current pricing table for available rates and account groups. Model IDs and prices come from TokenHot’s catalog; provider pricing and subscriptions are separate.
API
Code examples
curl --location --request POST 'https://api.tokenhot.ai/v1beta/models/nano-banana-2:generateContent' \
--header 'Authorization: Bearer <token>' \
--header 'Content-Type: application/json' \
--data-raw '{
"contents": [
{
"role": "user",
"parts": [
{
"text": "draw a baby"
}
]
}
],
"generationConfig": {
"responseModalities": [
"TEXT",
"IMAGE"
],
"imageConfig": {
"aspectRatio": "16:9",
"imageSize": "2k"
}
}
}'Blog
Related Articles
FAQ
Frequently asked questions
What is Nano Banana 2 best suited for?
Nano Banana 2 is suited to the input, output, and capability types listed on this page. Test production workloads before choosing it for a critical system.
How is Nano Banana 2 priced?
Pricing is shown in the pricing section above. Actual cost depends on usage volume, account group, and the request payload.
How can I call Nano Banana 2?
Use the model ID shown above with one of the supported API protocols. Code examples are provided when usage data is available.
Is Nano Banana 2 the same model as Nano Banana 2.1?
No. Google identifies Nano Banana 2 as Gemini 3.1 Flash Image and 2.1 as the updated model. Use the exact TokenHot route ID shown on the page you are integrating.
What endpoint does Nano Banana 2 use on TokenHot?
The TokenHot documentation shows POST /v1beta/models/nano-banana-2:generateContent with an image response modality.
Can I send a reference image?
Yes. The TokenHot Playground accepts one reference image per request. The published API example uses a text prompt, so confirm the image payload for your endpoint before automating edits.
How do I control the output format?
The TokenHot example shows aspectRatio and imageSize. Confirm the supported values and inspect output dimensions before committing to a channel workflow.
Are the scenario illustrations Nano Banana 2 outputs?
The illustrations were made with ImageGen to explain the workflows. Try the Playground or API with your own input to see how Nano Banana 2 performs.
Get started
Build with Nano Banana 2
API