Install Copilot
Find and install Copilot in Obsidian’s community plugin settings. Enable it according to your workspace policy, then open its model settings.Add a chat model
Find IDs in the model directory or the /v1/models response for your key. If a context limit is required, use the actual model parameters.
Test and select
Save and run the connection test. Open Copilot chat, select the new model, and send a short message. Then ask it to summarize a non-sensitive test note.Configure note retrieval
Chat and embedding models are configured separately. To create a vector index, configure a supported embedding model and endpoint. Successful chat does not establish a working indexing configuration. Rebuild the index when changing embedding models, as instructed by the plugin.Troubleshoot
If testing succeeds but chat fails, check the model selected in the chat panel. For large notes, check context and attachment limits. For 401 or 404, check key permissions and duplicated /v1 paths.Integration directory · Choose a model · Troubleshooting