Image Generation

Image generation

POST https://freemodel.in/v1/images/generations
POST https://freemodel.in/v1/images/edits
GET  https://freemodel.in/v1/images/models

Create images from a text prompt — or edit your own picture with a prompt — through the same OpenAI-compatible API you already use. It is the same engine as Image Studio in your dashboard, with the same models and the same price.

• Models: e.g. sunburst-2.5 — photoreal detail, up to 4K (3840 px). New image models appear by themselves in GET /v1/images/models and in Image Studio — no code change needed, just send the new model ID.
• Price: per image, paid from your account balance (sunburst-2.5: $0.0018; each model's price is in the models list). You pay only for images you actually receive — a failed request costs nothing.
• Works with: the official OpenAI SDKs (Python, Node), cURL and any tool that speaks the OpenAI Images API.

Quick start

curl https://freemodel.in/v1/images/generations \
  -H "Authorization: Bearer fm_your_key_here" \
  -H "Content-Type: application/json" \
  -d '{
    "model": "sunburst-2.5",
    "prompt": "A red fox sitting in fresh snow, golden hour, photoreal",
    "size": "1792x1024"
  }'

{
  "created": 1793981265,
  "data": [{ "b64_json": "iVBORw0KGgo…", "revised_prompt": "A red fox sitting in fresh snow…" }],
  "usage": { "images": 1, "requested": 1, "cost_usd": 0.0018, "tokens": 0 },
  "model": "sunburst-2.5",
  "size": "1792x1024",
  "id": "img_3f1c9a0b7d2e4c5a61"
}

List image models

No key needed. Returns every image model that is on right now, with its price per image, sizes and whether it can edit.

curl https://freemodel.in/v1/images/models

{
  "object": "list",
  "data": [{
    "id": "sunburst-2.5", "object": "model", "type": "image",
    "name": "Sunburst 2.5", "description": "Photoreal detail · up to 4K",
    "endpoints": ["/v1/images/generations", "/v1/images/edits"],
    "price_usd_per_image": {"1k": 0.0018, "2k": 0.0018, "4k": 0.0018},
    "sizes": ["1024x1024", "1792x1024", "3840x2160", "…"],
    "max_n": 4, "edit": true
  }]
}

Image models are listed here, not in /v1/models (that list is for chat models, so chat apps never pick an image model by mistake).

Request

Sizes

Pick a shape and a resolution. Every resolution costs the same.

Python

import base64
from openai import OpenAI

client = OpenAI(api_key="fm_your_key_here", base_url="https://freemodel.in/v1")

img = client.images.generate(
    model="sunburst-2.5",
    prompt="A cozy wooden cabin in snowy mountains at night, northern lights above",
    size="3840x2160",
    n=1,
)
with open("cabin.png", "wb") as f:
    f.write(base64.b64decode(img.data[0].b64_json))

Node.js

import OpenAI from "openai";
import fs from "node:fs";

const client = new OpenAI({ apiKey: "fm_your_key_here", baseURL: "https://freemodel.in/v1" });

const img = await client.images.generate({
  model: "sunburst-2.5",
  prompt: "A premium matte black water bottle on a stone pedestal, studio light",
  size: "1024x1024",
  response_format: "url",
});
console.log(img.data[0].url); // signed link, valid 7 days

Edit an image

Send 1–4 reference images with a prompt. The model looks at your images and follows the prompt — change colours, add or remove things, restyle, turn a photo into a poster.

curl https://freemodel.in/v1/images/edits \
  -H "Authorization: Bearer fm_your_key_here" \
  -F model=sunburst-2.5 \
  -F image=@car.jpg \
  -F prompt="Change the car colour to bright red, keep everything else the same" \
  -F size=1792x1024

img = client.images.edit(
    model="sunburst-2.5",
    image=open("car.jpg", "rb"),          # or a list of up to 4 files
    prompt="Turn this into a snowy winter night with a full moon",
    size="1792x1024",
)

References must be JPG, PNG or WEBP, up to 10 MB each. They are scaled to at most 2048 px before the model sees them.

Billing

• When a request starts, the price of the images you asked for is held from your balance. When it finishes you are charged only for the images you received; the rest goes straight back. A failed request is fully refunded.
• Parallel requests can never spend the same money twice — if your balance covers 3 images, at most 3 are created, however fast you send requests.
• Every answer carries the exact cost in usage.cost_usd and the headers X-FreeModel-Cost-Usd and X-FreeModel-Request-Id.
• Your images are stored on FreeModel: you also find them in Image Studio → History on any device.

Errors

Errors use the OpenAI shape: {"error": {"message", "type", "code"}}. The code is always one of ours:

Limits

• Up to 4 images per request and 4 requests running at the same time per account.
• A request takes about 25–45 seconds (4K takes the longest). Set your HTTP client timeout to at least 300 seconds.
• Prompts up to 4000 characters.