Image generation API (images/generations)

Generate and edit images with POST /v1/images/generations and the OpenAI SDK: automatic model picking, reference images, aspect ratio, URL or base64.

POST /v1/images/generations has OpenAI's shape, so client.images.generate works unchanged. The answer is synchronous: the request stays open until the image exists (usually seconds, at most about a minute).

Python

from openai import OpenAI

client = OpenAI(base_url="https://api.uttapen.ir/v1", api_key="sk-up-...")

img = client.images.generate(
    model="uttapen/auto-image",          # or a specific model, e.g. bytedance-seed/seedream-4.5
    prompt="a café poster with the words 'Morning Coffee' on a brick wall, warm light",
    n=1,
)
print(img.data[0].url)                    # a signed link, valid for 24 hours

Node

import OpenAI from "openai";

const client = new OpenAI({ baseURL: "https://api.uttapen.ir/v1", apiKey: "sk-up-..." });

const img = await client.images.generate({ model: "uttapen/auto-image", prompt: "a ceramic cup on a wooden table, soft light" });
console.log(img.data[0].url);

Editing from a reference image

Send one or more images (up to 4, 10 MB each) in image or images, as an https URL or a data:image/…;base64 URI. With uttapen/auto-image, an attached image routes the request to an editing model. A model without image input answers 400 with param: "images".

img = client.images.generate(
    model="uttapen/auto-image",
    prompt="make the background a beach at sunset",
    extra_body={"image": "https://example.com/product.jpg"},
)

Fields

Field	Meaning
`model`	a model id; empty or `uttapen/auto-image` means automatic picking
`prompt`	up to 4,000 characters, Persian or English
`n`	number of images, 1 to 4
`size` / `aspect_ratio` / `resolution` / `quality`	e.g. `1024x1024`, `16:9`, `2K`, `high`; each model supports a subset
`output_format`	`png`, `jpeg`, `webp`
`response_format`	`url` (default) or `b64_json`
`image` / `images`	reference images for an edit

Answer

{
  "created": 1791071431,
  "model": "bytedance-seed/seedream-4.5",
  "data": [{ "url": "https://api.uttapen.ir/v1/files/…/content?exp=…&sig=…", "mime_type": "image/png", "expires_at": "…" }],
  "uttapen": {
    "id": "01a1042d-…",
    "status": "completed",
    "cost_toman": "11605.000000",
    "hold_toman": "26186.439266",
    "requested_model": "uttapen/auto-image",
    "auto": { "intent": "photoreal", "reason": "…", "classifier": "llm" },
    "classifier_cost_toman": "11.227838"
  }
}

cost_toman is the final cost of this run; hold_toman is the hold that was released straight away. If your connection drops, the run still finishes and the result is available from GET /v1/uttapen/media/{id} (the id is in the X-Uttapen-Media-Id header).

Price

Each model's per-image price is on its page under image models. The final charge is what the model maker reports for that image; a failed run, or one that delivers no image, costs nothing. Billing and file retention are covered in the media overview.