The Custom provider allows setting up an OpenAI-API-compatible custom AI service. That could be an instance where you are self-hosting or have an external provider with such an offering. Depending on the service, you have to collect and fill in the following information:

• AI Connection Name: This name is always used when a specific AI connection is used.
• API URL (OpenAI compatible): The base URL of your OpenAI-compatible API endpoint (e.g. https://api.example.com/v1).
• Bearer Token: Optional API key or bearer token for authenticating requests to the custom endpoint.
• System Prompt: A preset system prompt to instruct the AI to answer with a specific behavior.
• Model: The optional model name that you want to access on your provider.
• Max Context Length (tokens): Maximum number of tokens (input and output combined) the model can process. Exceeding this may cause errors or truncation.