Connect Jan

Jan Desktop can use a remote API alongside its local models. This guide assumes Jan is installed. Copy the service Base URL above and keep your service API key ready.

1. Create a custom provider

Open Settings → Model Providers and click Add Provider (+). In Add Custom Provider, choose OpenAI-compatible.

Field	Value
Provider name	A unique name, such as my-api
Base URL	The service Base URL above, including /v1
API key	Your service API key, entered privately without Bearer

Select Create. Keep the Base URL at the API root; Jan adds /models/ for discovery and /chat/completions for text generation. Do not append either endpoint to the field.

2. Review the models

Jan tries to fetch models when you save. In the provider’s Models section, enable or add the text model you want. For manual entry, select + and use an exact ID returned by our authenticated models endpoint.

Jan does not automatically determine every custom model’s capabilities. Enable Tools only for a model with confirmed tool-calling support. Leave unverified vision and audio capabilities off. A model’s name alone does not establish its capabilities.

3. Start a new conversation

Return to chat and choose the saved custom provider and model in the model selector. Start a fresh conversation and send a short message; this generates a paid API request. Confirm it in your service dashboard.

The provider is saved in Jan Desktop. If its models do not appear, reopen the selector or restart Jan. A 404 commonly means the Base URL is missing /v1; a 401 or 403 calls for checking the key and model access. Do not replace existing local-model settings to troubleshoot a remote connection.

This follows Jan’s custom endpoint documentation. Jan Agent is a separate application with its own configuration; these desktop steps do not configure it.