Skip to main content
Last updated: 2026-09-18
Need the call to return immediately and fetch the image URL later? This model supports async image generation: swap the Gemini suffix for :asyncGenerateContent, or add a Prefer: respond-async header on the OpenAI endpoints — the request body stays byte-identical.

Nano Banana Pro

Use nano-banana-pro for image generation and reference-image editing through one public OpenAI Images API style contract.

What This Page Covers

  • This page only covers nano-banana-pro
  • The customer-facing entry point is POST /v1/images/generations
  • The input image field is image_input
  • The default public contract remains URL-based image_input; designated gray-release accounts additionally support data URLs, raw Base64, and three-image multipart editing
  • data URLs, Base64, and multipart are Crazyrouter compatibility extensions, not a claim that every upstream carrier natively supports them
  • Gemini / Vertex and other providers may sit behind this route, but those provider details are not part of the customer contract

Capability Status

The Pricing page and this doc use the same status language: Verified means there is a current minimal live artifact, while Beta means the official capability exists but the public validation set or edge coverage is still being completed.

Currently Validated Capabilities

The following conclusions come from historical live validation and the August 18, 2026 production gray release:
  • official Vertex channel 318 (vertexsophie) direct truth:
    • 2 reference images work
  • local relay on localhost:4000, pinned to channel 318, verified:
    • single WEBP URL reference works
    • single JPEG URL reference works
    • 2 PNG URL references work
    • resolution: "1K" works
    • resolution: "4K" works
    • aspect_ratio: "1:1" works
    • aspect_ratio: "21:9" works
    • output_format: "png" works
    • output_format: "jpg" works
    • the response returns data[].url
For designated gray user 11257 on production channel 304, we verified:
  • data URLs, raw Base64, and multiple image_input images on /v1/images/generations
  • multipart image, image[], and image[n] fields on /v1/images/edits
  • three-image ordering for original, annotation, and reference images
Do not present the following as fully promised customer-facing capability yet:
  • whether the exact multi-image limit is 8 or higher
  • 2K
  • exact large-file URL limit
  • exact 30 MiB boundary
  • additional aspect_ratio combinations
  • exact output_compression semantics
Non-gray accounts remain URL-only. Base64, data URLs, and multipart editing are currently enabled only for designated gray accounts.

Request Parameters

Current Multi-Image Status

  • 2 references: verified
  • gray user 11257 with three references: verified
  • higher counts: exact upstream max is still pending, so do not promise it publicly yet

Routing Note

  • single-image requests can use normal available carriers
  • multi-image requests normally use official Gemini / Vertex carriers; designated gray accounts may use an explicitly verified gray carrier
  • that is a server-side routing rule; the customer contract stays the same

Gray three-image multipart editing

Gray user 11257 can use /v1/images/edits:
Images are ordered by submission order. The annotation image is treated as a visual reference; pixel-perfect GPT-mask semantics are not guaranteed.

Request Examples

Single-Image Edit

Two-Reference Input


Response Example

The recommended automation path is to consume data[].url:
To retrieve the generated image, read data[0].url. Crazyrouter currently archives upstream inline image output and returns a public URL to the client.

Validation Snapshot

Current Known Limits

  • Non-gray accounts cannot use image, data: / Base64, or multipart image input
  • The gray path is verified up to three images; higher limits are not promised yet
  • the docs do not promise the exact 30 MiB URL boundary yet
  • the docs do not promise full closure on every screenshot-listed ratio or parameter yet
If you need Gemini native generateContent image endpoints, see Gemini Image Generation and Gemini Image Editing. If you care more about reference-count breadth, also see Nano Banana and Nano Banana 2.