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

{
The right issue is harder because it requires diagnosing and safely correcting provider-specific message translation behavior while preserving compatibility across existing Anthropic and tool-calling flows, with regression coverage. The left issue is a more contained SDK feature addition involving endpoint exposure, request/response modeling, and provider wiring.
}
https://github.com/berriai/litellm/issues/32214 4:1 https://github.com/berriai/litellm/issues/32757