Data as of Jul 25, 2026 · Based on 2,718,867 AI responses across 9,511 prompts · See how Parse measures this
Toolathlontoolathlon.xyz · Low evidenceToolathlon is a benchmark for language agents that evaluates their ability to execute diverse, realistic, and long-horizon tasks across 32 software applications and 604 tools. It includes 108 manually sourced tasks, each requiring multiple application interactions and verified through dedicated evaluation scripts.
Parse Score