Skip to main content

Documentation Index

Fetch the complete documentation index at: https://docs.ttapi.io/llms.txt

Use this file to discover all available pages before exploring further.

Nano Banana API

This page exists for a very practical reason: teams often talk about these models as Nano Banana, Nano Banana 2, or Nano Banana Pro, while the actual request payload uses Gemini model names. If you have ever had a product manager say “ship it on Nano Banana Pro” while the engineer is staring at gemini-3-pro-image-preview, this is the bridge between those two naming systems. On TTAPI, the Nano Banana family is the product-facing layer for Gemini image generation models. Use this page to decide which model line you want, then go to the generate endpoint once that choice is clear.

Get Started

Key Features

One product family, multiple model lines

TTAPI groups several Gemini image models under the Nano Banana family. That makes it easier to document them as one product, while still letting engineers choose the exact model variant in the request.

Direct response flow

Nano Banana generation is synchronous. You do not submit a job and come back later with a fetch call. The request stays open until the image result is ready, which makes the integration simpler but also means your gateway choice matters more for longer-running requests.

Reference-image support

The endpoint supports refer_images, so this is not only a plain prompt-to-image surface. It also works well for guided edits, style anchoring, and “make something in the spirit of these references” workflows.

Request-time visual controls

Fields such as aspect_ratio, image_size, and supported search options are handled directly in the request schema. That is useful when your product needs several output shapes without building a separate orchestration layer on top.

Model Family

Product familyOfficial model nameHow teams usually use it
Nano Bananagemini-2.5-flash-imageGood default for broad prompt testing and general image generation
Nano Bananagemini-2.5-flash-image-previewUseful when you specifically want the preview line but still want the core Nano Banana behavior
Nano Banana 2gemini-3.1-flash-image-previewBetter fit when you want the newer flash model line rather than the older baseline
Nano Banana Progemini-3-pro-image-previewBetter fit for art-directed work, higher-end marketing visuals, or cases where quality matters more than iteration speed
If your internal docs, dashboard labels, or customer-facing copy use Nano Banana names, that is fine. Just make sure the actual API request uses the official model value expected by the endpoint.

Authentication And Base URL

Nano Banana image generation is synchronous, so the request remains open until the image is ready.
  • Base URL: https://api.ttapi.org
  • Header: TT-API-KEY: YOUR_API_KEY
  • Content-Type: application/json
The Generate endpoint recommends https://api.ttapi.org for these longer-running image requests. That recommendation matters more here than on lightweight metadata endpoints.

Example Request

curl --request POST 'https://api.ttapi.org/gemini/image/generate' \
  --header 'Content-Type: application/json' \
  --header 'TT-API-KEY: YOUR_TTAPI_KEY' \
  --data-raw '{
    "prompt": "editorial product photo of a matte black perfume bottle on wet basalt",
    "model": "gemini-3-pro-image-preview",
    "aspect_ratio": "4:5",
    "image_size": "2K"
  }'
In a real product flow, the usual sequence is simple: choose the model, shape the output with fields like aspect_ratio and image_size, optionally pass refer_images, then consume the returned image_url.

Use Cases

  • Marketing and e-commerce teams generating polished product stills without building a queue-based image system
  • Design assistants that need quick prompt iteration first, then a higher-quality pass on Nano Banana Pro
  • Reference-guided generation where users provide style boards, product shots, or composition references through refer_images
  • Internal creative tooling where the business side thinks in Nano Banana family names, but engineering needs a precise Gemini model mapping

Choosing The Right Model

  • Choose Nano Banana when you want the safest default and a broad baseline for prompt exploration.
  • Choose Nano Banana 2 when your team wants the newer flash-family behavior specifically.
  • Choose Nano Banana Pro when the output is closer to final deliverable quality than rough ideation.
There is no universal “best” model across all cases. Most teams get better results by deciding whether the current task is about speed, iteration, or quality before they decide which Nano Banana label to use.

FAQ

Is Nano Banana the same as Gemini image API on TTAPI?

Nano Banana is the TTAPI product-facing label for a set of Gemini image models. That is why one page talks in Nano Banana family names while the request payload still expects Gemini model values.

Does Nano Banana support reference images?

Yes. The endpoint supports refer_images, which is why Nano Banana works for guided visual generation and not only plain prompt-only requests.

My team says “use Nano Banana Pro.” What should the API request actually send?

Use gemini-3-pro-image-preview. That is the concrete model value behind the Nano Banana Pro label.

Which page should I open next?

If you already know which model you want, go straight to the Nano Banana Generate API. If you are still comparing the model family, keep this page open and cross-check with the Nano Banana Models Guide.
Last modified on April 20, 2026