Image Generation

Create the .prompty File

Set apiType: image and pick a DALL-E model. Image options go in model.options.additionalProperties:

image-gen.prompty

---
name: image-gen
description: Generate images from text descriptions
model:
  id: dall-e-3
  provider: openai
  apiType: image
  connection:
    kind: key
    apiKey: ${env:OPENAI_API_KEY}
  options:
    additionalProperties:
      size: "1024x1024"
      quality: standard
      style: vivid
inputs:
  - name: prompt
    kind: string
    default: A serene mountain landscape at sunset
---
{{prompt}}

Note

No role markers needed — the rendered text is sent directly as the image prompt.

---

Generate an Image

invoke() returns the generated image URL:

Python

from prompty import invoke

url = invoke("image-gen.prompty", inputs={
    "prompt": "A cat astronaut floating in space, digital art"
})
print(url)  # https://oaidalleapiprodscus.blob.core.windows.net/...

TypeScript

import { invoke } from "@prompty/core";
import "@prompty/openai";

const url = await invoke("image-gen.prompty", {
  prompt: "A cat astronaut floating in space, digital art",
});
console.log(url);

C#

using Prompty.Core;

var url = await Pipeline.InvokeAsync("image-gen.prompty",
    new() { ["prompt"] = "A cat astronaut floating in space, digital art" });
Console.WriteLine(url);

Rust

Caution

The Rust runtime does not yet support apiType: image. This feature is planned for a future release.

Caution

Image URLs are temporary(≈ 1 hour). Download anything you want to keep.

---

Configuration Options

All options live under model.options.additionalProperties:

DALL-E 3

Option	Values	Default
size	1024x1024, 1792x1024, 1024x1792	1024x1024
quality	standard, hd	standard
style	vivid, natural	vivid

DALL-E 2

Option	Values	Default
size	256x256, 512x512, 1024x1024	1024x1024
n	1–10	1

Tip

Use dall-e-2 with size: "256x256" during development — much cheaper and faster for iteration.

---

Azure OpenAI

Same pattern — switch provider and connection:

image-gen-azure.prompty

---
name: azure-image-gen
model:
  id: ${env:AZURE_OPENAI_IMAGE_DEPLOYMENT}
  provider: azure
  apiType: image
  connection:
    kind: key
    endpoint: ${env:AZURE_OPENAI_ENDPOINT}
    apiKey: ${env:AZURE_OPENAI_API_KEY}
  options:
    additionalProperties:
      size: "1024x1024"
      quality: standard
inputs:
  - name: prompt
    kind: string
---
{{prompt}}

The calling code is identical — only the .prompty file changes.

---

Further Reading

• File format reference — full .prompty frontmatter syntax
• Connections — configuring provider connections