Connect DeepChat

This guide is for the DeepChat desktop app by ThinkInAIXYZ. Have the app installed, your service key ready and an exact model ID available from our models endpoint.

1. Add the connection

Open DeepChat’s settings and its model provider page. Select Add Custom Provider and fill in:

Field	Value
Name	my-api, or another recognizable local name
API Type	OpenAI Completions
API Base URL	The service Base URL above, including /v1
API Key	Your service key, pasted privately without Bearer

In the reviewed version, OpenAI Completions means OpenAI-compatible Chat Completions. The separate OpenAI option uses Responses. Confirm the preview ends in /v1/chat/completions; do not enter the complete request path in the Base URL.

Choose Save without testing to persist the connection first, or Connect and load models to verify it and discover models. Connection checks may send a text-model request and spend balance. A successful check applies to the model tested, rather than every listed model.

2. Enable a text model

Open the provider’s Models tab. Use Refresh Models if needed, then enable the text model you want. Check its ID against the service’s authenticated model list. Discovery and enabling are separate steps.

Select Start chatting, or open a fresh chat and choose the saved provider and enabled model in the model picker. The app saves the connection. Reopen the picker or restart DeepChat if it still shows an older list.

3. Verify and troubleshoot

Send a short text prompt and confirm the model and request in your service dashboard. If no models appear, check the URL, key and enable switches. For 404, look for a missing or duplicated /v1. Tool use and other agent features require separate capability checks and permissions.

These instructions follow the reviewed custom provider form and English UI labels. Older versions may use different wording.