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

{
The left issue spans provider-specific request translation, asynchronous or conditional cost calculation, and propagation through response metadata, callbacks, storage, and UI, making diagnosis and end-to-end validation broader. The right issue is comparatively localized to callback metadata modeling and credential isolation, with focused code changes and tests.
}
https://github.com/berriai/litellm/issues/30725 3:1 https://github.com/berriai/litellm/issues/30825