Rather than testing one narrow tool-calling skill, Tool Decathlon spreads its evaluation across ten varied tasks to check whether an agent's tool-use ability generalizes rather than being tuned for a single benchmark's tool set.
Tool Decathlon is a benchmark that scores an AI agent across ten distinct tool-use tasks spanning different tool types and domains.
Rather than testing one narrow tool-calling skill, Tool Decathlon spreads its evaluation across ten varied tasks to check whether an agent's tool-use ability generalizes rather than being tuned for a single benchmark's tool set.