The gptel-litellm module, which depends on the uuidgen library, adds tracking of “"sessions"” for users with a LiteLLM proxy backend — where each GPTel Chat buffer constitutes its own session.
What this means is that all requests from the same buffer are grouped under the same session-id in LiteLLM’s interface, for accounting and cost tracking purposes. An example of what this looks like can be seen in the documentation.
I also recommend setting a tag for GPTel, so LiteLLM can see all requests that orginated from GPTel no matter which buffer was used:
(gptel-make-openai ""LiteLLM""
:key gptel-api-key
:host ...
:models ...
:header
(lambda () (when-let* ((key (gptel--get-api-key)))
`((""x-api-key"" . ,key)
(""x-litellm-tags"" . ""gptel"")))))