Worked on expanding the capabilities of the BerriAI/litellm repository by integrating the Kimi K2.5 model from OpenRouter, enabling vision features, function calling, and detailed pricing-aware usage. Focused on API integration and data validation using Python and JSON, the work introduced guidance for pricing and token limits to support new business workflows. Testing was employed to ensure robust model integration and future extensibility. This feature laid the foundation for broader AI-enabled workflows by aligning technical enhancements with business goals, while establishing patterns for pricing-aware usage and setting the stage for future improvements in the litellm codebase.