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

{
The left issue is harder because it requires extending cross-provider request translation for multimodal content, preserving semantics across differing schemas, and adding coverage for multiple lowering paths. The right issue is a localized precedence correction across a few provider branches with comparatively straightforward regression tests.
}
https://github.com/berriai/litellm/issues/28232 5:1 https://github.com/berriai/litellm/issues/26170