Use a local LLM proxy to aggregate AI models

Describe the feature you'd like to request

Allow administrator to set up access to multiple LLMs in nextcloud via proxy

Describe the solution you'd like

LITELLM does an incredible job of this.

Combination in Nextcloud would allow more options than the current OpenAI, LocalAI and Replicate options but would also provide a single AI abstraction layer and allow integration with Nextcloud user management, etc

LiteLLM's model aliases provide a useful way to have and call one proxy with unlimited Apis beneath.

The integration could allow integration with Nextcloud group permissions and tasks, allowing other apps to call the various AI models via alias and permissions through the abstraction layer.

If the aliases are exposed as bots then they could be called in Talk, etc

This dovetails with the following Nextcloud Assistant enhancement request - https://github.com/nextcloud/assistant/issues/76

Describe alternatives you've considered

Currently calling LiteLLM via the Nextcloud LocalAI integration

No files, tests, or entry points are named. Start by reviewing the existing Nextcloud LocalAI integration and the linked Assistant enhancement request; done would require an agreed design for LiteLLM proxy configuration, aliases, permissions, and app integrations.