Image Generation

Method	Path
`POST`	`/v1/images/generations`

The request body follows the OpenAI Images-compatible shape and is forwarded to an upstream provider that supports image generation. The main entrypoint is POST /v1/images/generations.

bash

curl https://api.modelgo.ai/v1/images/generations \
  -H "Authorization: Bearer $MODELGO_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{"model":"<model>","prompt":"a cat in neon city"}'

Common optional fields include size, quality, and n, subject to upstream model support.

Response shape

Whether data[] carries url or b64_json follows each model's official behavior: the gateway does not convert between the two and does not host images on your behalf. response_format is forwarded to models that accept it; for models that do not, see the table below.

Model	Default shape	`response_format`
gpt-image family (`gpt-image-2` / `gpt-image-2.5-*`)	`b64_json`	Not accepted upstream; the gateway strips it before forwarding without an error, and always returns `b64_json`
Qwen Image (`qwen-image-3.0` / `qwen-image-3.0-pro`)	`url` (link valid for 24 hours)	`b64_json` does not error but is ignored; the response still carries a `url`
Seedream (`doubao-seedream-*`)	`url` (link valid for 24 hours)	Both `url` and `b64_json` are supported

When switching models, do not hard-code data[0].url or data[0].b64_json; read whichever field is present (example: Image editing). Per-model details: GPT Image, Qwen Image, Seedream.

Seedream size limit example

Models such as doubao-seedream-5.0-lite impose a lower bound on size; too small a value is rejected upstream (HTTP 400 provider_invalid). For example, 1024x1024 may fail while 2048x2048 succeeds. See the official documentation for the exact minimum; full curl: API catalog.