Connect an OpenAI-compatible endpoint (hosted or self-hosted model)
Point DashboardGPT at any server that speaks the OpenAI chat API: vLLM, Ollama, LM Studio, LiteLLM or a hosted gateway.
For the complete AI-readable content index, see /llms.txt.
Settings & preferences · 1 min read
- 1Open Settings → AI, choose “Add provider” and pick “OpenAI compatible” as the Provider.
- 2Enter the Base URL of the API (usually ending in /v1, for example http://host:11434/v1 for Ollama) and the Model name exactly as the endpoint serves it.
- 3Add an API key only if the endpoint requires one; leave it empty for an open endpoint.
- 4Choose “Test connection”. DashboardGPT asks the endpoint for its model list and asks the named model for one token, so a wrong URL or model name is caught here.
- 5Save. You can re-test any time with “Test” on the provider card.
Related reading
Guides from Learn on the why and when behind this.
Still stuck?
Ask the docs assistant about “Connect an OpenAI-compatible endpoint (hosted or self-hosted model)” or anything else in DashboardGPT.