OpenAI-compatible providers

Follow the steps below to configure the AI Autodocumentation feature in Dataedo with any service that implements the OpenAI chat completions protocol, such as Groq, OpenRouter, vLLM, Ollama, or LM Studio.

Prerequisites

caution

To use the AI Autodocumentation feature with an OpenAI-compatible provider, you need the endpoint of its API, the model ID, and an API key. These are required to access the provider's API and use its models for autodocumentation.

You will also need the Admin role in Dataedo Portal to add the engine, and (Power) Data Steward permissions to the Dataedo Repository to create AI-generated descriptions.

Provider details

Get the endpoint, model ID, and API key

• Endpoint: the base URL of the API, without the /chat/completions path. See the examples below.
• Model ID: the model ID exactly as the service expects it, for example, llama-3.3-70b-versatile on Groq or llama3.1:8b on Ollama.
• API key: the key issued by the service. If your self-hosted server does not check API keys, use any placeholder value, as the field is required in Dataedo.

Service	Example endpoint
Groq	https://api.groq.com/openai/v1/
OpenRouter	https://openrouter.ai/api/v1/
vLLM	http://<server>:8000/v1/
Ollama	http://<server>:11434/v1/
LM Studio	http://<server>:1234/v1/

caution

The endpoint must be reachable from the Dataedo Portal server. localhost refers to the machine that runs Dataedo Portal, not to your computer.

Description quality depends on the model. Small local models may return incomplete or generic descriptions.

Configure AI autodocumentation in Dataedo

1. In Dataedo navigate to Settings > System Settings > LLM (AI) engines.
2. Click + Add engine.
3. In the New engine window, select OpenAI-compatible.
4. Fill in the engine details:
   • Title [A]: a user-friendly name shown to users when they choose an engine, for example Llama 3.3 (Groq).
   • Endpoint [B]: enter the endpoint of the API, for example https://api.groq.com/openai/v1/.
   • Model [C]: enter the model ID. There is no model list for OpenAI-compatible providers.
   • API key [D]: paste the API key issued by the service.
5. (Optional) Expand Advanced settings [E] to change Max tokens (default 4096) or to add Additional context for the prompts. See Add an AI engine for details.
6. Click Save and test.

Dataedo Portal sends a short test prompt to the provider and saves the engine only if it responds. If the test fails, check the error message, correct the settings, and click Save and test again.