Sorting LiteLLM GitHub issues by difficulty to implement. Higher = harder.

{
The right item is harder because it introduces a cross-cutting quota capability requiring configuration design, time-window accounting, shared or persistent state, concurrency handling, routing behavior, and broad testing. The left item appears comparatively localized to scheduler selection logic with focused regression coverage.
}
https://github.com/berriai/litellm/issues/14398 5:1 https://github.com/berriai/litellm/issues/25323