September 2026
Text-to-World Bench
Can a text become a completed real-world task? Testing text agents on the bookings and purchases they promise to handle.
7 agents · 11 tasks · 4 task worlds
Read the write-upResults
Task success rate
Hover or select an agent to see its task-world breakdown and latency.
For more information about the task set, contact info@rundelphi.com.