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

{
The right issue is harder because it requires tracing shared sync and async request flows across multiple provider adapters, preserving timeout propagation through transformation layers, and adding regression coverage without disrupting native API paths. The left issue is comparatively localized model registration and compatibility testing.
}
https://github.com/berriai/litellm/issues/28132 4:1 https://github.com/berriai/litellm/issues/30754