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

{
Adding a multi-modal provider/model integration spanning video, image, and audio requires broader API mapping, validation, capability handling, and regression coverage across several request and response paths. The cache directive bug is comparatively localized to one transformation path with a focused test and fix.
}
https://github.com/berriai/litellm/issues/16073 6:1 https://github.com/berriai/litellm/issues/27950