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

{
The Bayesian optimization request is substantially broader: it requires designing and integrating an experimentation and optimization system, defining objective and configuration interfaces, managing trials and persistence, and handling operational cost and safety concerns. The other item is a focused compatibility/debugging effort within an existing request transformation and caching pipeline, though reproducing the multi-turn behavior adds some risk.
}
https://github.com/berriai/litellm/issues/30448 4:1 https://github.com/berriai/litellm/issues/29810