Is it agentic enough? Benchmarking open models on your own tooling

Researchers developed a benchmark to evaluate open AI models on their ability to use external tools effectively. This framework helps assess how well models integrate with user-provided tooling for enhanced agentic behavior.