> ## Documentation Index
> Fetch the complete documentation index at: https://docs.collabase.ch/llms.txt
> Use this file to discover all available pages before exploring further.

# OpenAI

> Generate text with GPT models, create images with DALL·E 3, transcribe audio, convert text to speech, and generate embeddings.

# OpenAI

Use OpenAI's API directly in automations — run GPT-4o completions, generate images with DALL·E 3, transcribe audio files, produce speech, and create embeddings for semantic search pipelines.

## Authentication

**Auth type:** API Key

Create a secret key at [platform.openai.com/api-keys](https://platform.openai.com/api-keys).

| Credential          | Description                                                   |
| ------------------- | ------------------------------------------------------------- |
| **API Key**         | Your OpenAI secret key (starts with `sk-`)                    |
| **Organization ID** | Optional — found in OpenAI Platform → Settings → Organization |

## Actions

<AccordionGroup>
  <Accordion title="Text">
    | Action              | Key inputs                                                                     | Key outputs                                                 |
    | ------------------- | ------------------------------------------------------------------------------ | ----------------------------------------------------------- |
    | **Chat Completion** | `model`, `systemPrompt`, `userMessage`, `temperature`, `maxTokens`, `jsonMode` | content, model, promptTokens, completionTokens, totalTokens |
  </Accordion>

  <Accordion title="Images">
    | Action             | Key inputs                                                        | Key outputs             |
    | ------------------ | ----------------------------------------------------------------- | ----------------------- |
    | **Generate Image** | `prompt`, `model` (dall-e-3/dall-e-2), `size`, `quality`, `style` | imageUrl, revisedPrompt |
  </Accordion>

  <Accordion title="Audio">
    | Action               | Key inputs                                                              | Key outputs           |
    | -------------------- | ----------------------------------------------------------------------- | --------------------- |
    | **Transcribe Audio** | `audioUrl`, `language`, `prompt`                                        | text, language        |
    | **Text to Speech**   | `input`, `voice` (alloy/echo/fable/onyx/nova/shimmer), `model`, `speed` | audioBase64, mimeType |
  </Accordion>

  <Accordion title="Embeddings & moderation">
    | Action               | Key inputs                                        | Key outputs                     |
    | -------------------- | ------------------------------------------------- | ------------------------------- |
    | **Create Embedding** | `input`, `model` (text-embedding-3-small / large) | embedding\[], dimensions, model |
    | **Moderate Content** | `input`                                           | flagged, categories{}, scores{} |
    | **List Models**      | —                                                 | models\[] — id, created         |
  </Accordion>
</AccordionGroup>

<Tip>
  Enable `jsonMode: true` in **Chat Completion** to force GPT to return valid JSON output. Your `systemPrompt` or `userMessage` must instruct the model to produce JSON — the API enforces the format but does not define the schema.
</Tip>