Developed Qwen2.5 model support for the sgl-project/mini-sglang repository by designing a new model class and integrating it into the existing model registry. This approach enabled runtime selection of Qwen2.5, expanding the platform’s compatibility with emerging deep learning models. The work established an extensible registry pattern, simplifying the process of adding future models with minimal code changes. Leveraging Python and expertise in machine learning and model development, the implementation laid the foundation for broader inference model support. The contribution addressed the need for scalable model integration, aligning with the project’s roadmap for a more flexible and maintainable architecture.