Skip to main content
MarkdownLLMs.txt

Nano Banana 2.1 — Gemini generateContent

Overview​

Use gemini-nano-banana-2.1 to generate an image from text or edit up to 14 reference images with the Gemini request format. Each synchronous request returns one image as Base64 in an inlineData response part.

Endpoint​

POST /v1beta/models/gemini-nano-banana-2.1:generateContent

The base URL is https://api.uniall.ai. The main examples use Google's official model ID gemini-nano-banana-2.1. The compatible URL /v1beta/models/nano-banana-2.1:generateContent also remains available.

Authentication​

Use your UniAI user API key:

x-goog-api-key: <UNIAI_API_KEY>
Content-Type: application/json

Request Parameters​

Send one user entry in contents. Use camelCase field names.

ParameterTypeRequiredDescription
contentsarrayYesExactly one user-content entry.
contents[].rolestringYesuser.
contents[].parts[].textstringYesNon-empty generation or editing instruction.
contents[].parts[].inlineDataobjectConditionalReference image for editing; add a part for each image, up to 14.
contents[].parts[].inlineData.mimeTypestringConditionalimage/png, image/jpeg, or image/webp.
contents[].parts[].inlineData.datastringConditionalRaw Base64 image bytes without a Data URL prefix.
generationConfig.imageConfig.imageSizestringNo1K, 2K, or 4K; defaults to 1K.
generationConfig.imageConfig.aspectRatiostringNoOne of the 14 ratios below. If omitted, the ratio follows the reference image or generated content.
generationConfig.thinkingConfig.thinkingLevelstringNoMINIMAL, MEDIUM, or HIGH; defaults to MEDIUM.
generationConfig.responseModalitiesstring[]NoDefaults to ["IMAGE"]; also accepts ["TEXT", "IMAGE"]. Text appears only if final text is actually generated. ["TEXT"] alone is unsupported.
generationConfig.candidateCountintegerNoOnly 1 is accepted; each request returns one image.

Resolution And Reference Images​

Supported aspect ratios:

1:1  1:4  1:8  2:3  3:2  3:4  4:1
4:3  4:5  5:4  8:1  9:16  16:9  21:9

Resolution is a tier, not a fixed longest-edge pixel count. For example, a non-square 1K result can be 1376 × 768. 0.5K, 512, and 8K are unsupported.

Editing requires 1 to 14 PNG, JPEG, or WebP reference images. Keep them in parts in the order described by your prompt. Order and duplicate entries are preserved; duplicates count toward the limit. Set mimeType to the actual image type and put only raw Base64 in data.

Request Examples​

Text To Image​

curl -X POST "https://api.uniall.ai/v1beta/models/gemini-nano-banana-2.1:generateContent" \
  -H "x-goog-api-key: <UNIAI_API_KEY>" \
  -H "Content-Type: application/json" \
  -d '{
    "contents": [
      {
        "role": "user",
        "parts": [
          {"text": "Create a spring latte poster with a cream background and clear typography"}
        ]
      }
    ],
    "generationConfig": {
      "responseModalities": ["IMAGE"],
      "imageConfig": {"imageSize": "2K", "aspectRatio": "16:9"},
      "thinkingConfig": {"thinkingLevel": "HIGH"},
      "candidateCount": 1
    }
  }'

Image Editing​

Replace <BASE64_IMAGE_BYTES> with the reference image's raw Base64 bytes before sending this request.

curl -X POST "https://api.uniall.ai/v1beta/models/gemini-nano-banana-2.1:generateContent" \
  -H "x-goog-api-key: <UNIAI_API_KEY>" \
  -H "Content-Type: application/json" \
  -d '{
    "contents": [
      {
        "role": "user",
        "parts": [
          {"text": "Keep the cup shape and color; replace the background with a bright cafe"},
          {
            "inlineData": {
              "mimeType": "image/png",
              "data": "<BASE64_IMAGE_BYTES>"
            }
          }
        ]
      }
    ],
    "generationConfig": {
      "responseModalities": ["IMAGE"],
      "imageConfig": {"imageSize": "2K", "aspectRatio": "4:5"},
      "thinkingConfig": {"thinkingLevel": "MEDIUM"},
      "candidateCount": 1
    }
  }'

Add another inlineData part for each additional reference image. To make a later edit, use the previous output image as a reference in a new independent request.

Response Example​

{
  "candidates": [
    {
      "content": {
        "role": "model",
        "parts": [
          {
            "inlineData": {
              "mimeType": "image/png",
              "data": "<BASE64_IMAGE_BYTES>"
            }
          }
        ]
      },
      "finishReason": "STOP"
    }
  ]
}
Response fieldDescription
candidates[].content.parts[].inlineDataFinal image. Base64-decode data and save it according to the returned mimeType.
candidates[].content.parts[].textOptional final text when requesting ["TEXT", "IMAGE"] and final text is actually generated. Text is not guaranteed.
usageMetadataMay be present when actual usage metadata is available; do not require it to read the image.

Find the part containing inlineData rather than assuming the first part is an image. The response is synchronous and has no task ID or polling step. A client timeout does not confirm that generation stopped; avoid automatically submitting the request again.

Capability Limits And Common Errors​

  • Search is unavailable. Omit tools, enable_web_search, and enable_image_search; supplying them is rejected even when a search flag is false.
  • Only a single user-content entry with text and image references is supported. Multi-turn history, thinking signatures or summaries, video/PDF input, Files, Interactions, and Batch APIs are unavailable in this image integration.
  • quality, custom pixel size or custom_size, masks, transparent backgrounds, seed, and streaming output are unsupported.
  • Do not use /v1/chat/completions or /v1/images/tasks for Nano Banana 2.1.
  • An editing request without a reference image, more than 14 references, a Data URL prefix inside inlineData.data, an unsupported image size, text-only output, or a candidateCount other than 1 is invalid.
  • Thinking control is available only for the new 2.1 integration. It is not enabled for nano-banana-2, nano-banana-pro, or fixed-resolution compatibility models.