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

{
Supporting an additional multimodal content variant in shared token-counting and trimming paths carries broader compatibility and testing risk across providers and payload formats, while the other issue is a narrowly localized timezone-normalization correction with focused regression coverage.
}
https://github.com/berriai/litellm/issues/28409 3:1 https://github.com/berriai/litellm/issues/34896