AgentRealTaskBench
中文 Request a test

Task set v1.0 / Local: US apps

ARTB-G05 · v1.0

A hotel near the venue

Two nights near a conference venue, under a budget, with free cancellation.

Travel & placesTravelRecommendations

What we say to the agent

Find 3 hotels within 1 km of venue for dates, under budget/night, with free cancellation.

How it is judged

Pass
  • All 3 exist and meet distance, price and cancellation constraints.
Partial
  • 2 of 3 meet every constraint.
Fail
  • Fewer than 2, or an invented hotel.

Changes every round

  • {venue}
  • {dates}
  • {budget}

Prompt templates are public. The concrete values are published with each round's results.

Evidence we keep

  • screen_recording
  • verification_notes