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

{
The left issue is harder because it spans asynchronous streaming, provider-specific response normalization, mid-stream error handling, and compatibility across multiple model variants. The right issue is narrower: extending an existing guardrail payload transformation to support another content modality, with focused validation and tests.
}
https://github.com/berriai/litellm/issues/27928 3:1 https://github.com/berriai/litellm/issues/35332