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

{
The left issue is harder because it requires tracing a request option through proxy validation, provider-specific parameter translation, reasoning-generation behavior, and response normalization while preserving compatibility across models. The right issue is comparatively localized to an iterator’s accumulation strategy, with focused regression and performance tests.
}
https://github.com/berriai/litellm/issues/26413 4:1 https://github.com/berriai/litellm/issues/31861