​
Image Series

Image capabilities are currently mainly divided into six categories: the official Gemini image generation entry point, the GPT Image generation entry point, Jimeng Image (OpenAI-compatible and official formats), the unified Doubao Seedream image generation entry point, the OpenAI Images-compatible generation/editing entry point, and the dedicated Midjourney task interface.

Model Support Matrix

View image API support scope, reference image input methods, and ratio and size constraints by model family.

Gemini Image Generation

POST /v1beta/models/{model}:generateContent, covering gemini-3-pro-image-preview, gemini-2.5-flash-image-preview, and gemini-3.1-flash-image-preview in a unified way.

gpt-image-2

POST /v1/images/generations, calling gpt-image-2 according to basic ratio and size tiers.

gpt-image-2-pro

POST /v1/images/generations, with additional 2K and 4K high-resolution tiers.

Jimeng Image

OpenAI-compatible formats (generation/editing) and official formats (image generation/query), supporting jimeng-4.0 and jimeng-4.5.

Doubao Seedream

POST /v1/images/generations, supporting Seedream text-to-image and image array image-to-image.

OpenAI Images Compatible

Compatibility entry point descriptions based on POST /v1/images/generations, POST /v1/images/edits, and POST /v1/edits.

Midjourney

/mj/submit/*, /mj/task/*, /mj/image/{id}, suitable for imagine, blend, describe, upscale, and variation operations.

Upload and Result Query

First upload assets with /api/upload/presign, then pass the public URL into image generation, image editing, and video APIs.

​
Route List

MethodPathDescription
POST/v1/images/generationsUnified image generation entry point
POST/v1/images/editsOpenAI Images-compatible editing entry point
POST/v1/editsLegacy edit alias
POST/v1/images/variationsRoute reserved, not implemented yet
POST/jmJimeng official-format image generation/query entry point
POST/mj/submit/imagineMidjourney imagine
POST/mj/submit/blendMidjourney blend
POST/mj/submit/describeMidjourney describe
POST/mj/submit/editsMidjourney editing-related actions
POST/mj/submit/videoMidjourney video-related actions
GET/mj/task/{id}/fetchMidjourney task query
GET/mj/image/{id}Midjourney image proxy
POST/v1beta/models/{model}:generateContentGemini official-format image generation entry point
Although OpenAI-compatible paths such as /v1/images/variations, /v1/files*, and fine-tunes* are reserved in the routes, they currently return a not implemented error and cannot be used directly in production.
  1. Image Model Support Matrix
  2. Gemini Image Generation API
  3. gpt-image-2 Image Generation API
  4. gpt-image-2-pro Image Generation API
  5. Jimeng Image Overview
  6. Doubao Seedream Image Generation API
  7. OpenAI Images Compatible Overview
  8. OpenAI Images Compatible Image Editing
  9. Upload Images
  10. Midjourney Overview