Kaggle Launches Community Benchmarks for Custom AI Model Evaluation

Summary

On January 15, 2026, Kaggle launched Community Benchmarks, a new platform feature that enables developers to design, run, and share custom benchmarks for evaluating AI models. The tool allows developers to move beyond static academic metrics and build reproducible evaluations tailored to specific use cases. Key capabilities include custom task construction for code execution, tool use, and multi-turn conversations using a new kaggle-benchmarks SDK; free access to run benchmarks against state-of-