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

{
Issue 33987 is harder because it spans asynchronous batch polling, terminal-state handling, logging-worker error propagation, provider-specific file semantics, and zero-cost accounting, requiring broader regression coverage. Issue 28239 is comparatively localized to request-parameter normalization for embedding calls and vLLM compatibility.
}
https://github.com/berriai/litellm/issues/33987 4:1 https://github.com/berriai/litellm/issues/28239