II.
Benchmark overview
Reference · livebenchmark:appworld
AppWorld overview
Tool-use benchmark with a controllable ecosystem of 9 simulated apps (email, calendar, shopping, etc.) and 750 cross-app tasks. Tests long-horizon multi-app orchestration.
Attributes
displayName
AppWorld
homepageUrl
kind
tool-use
targetsKind
AgentVersion
description
Tool-use benchmark with a controllable ecosystem of 9 simulated
apps (email, calendar, shopping, etc.) and 750 cross-app tasks.
Tests long-horizon multi-app orchestration.
Outgoing edges
applies_to1
- domain:software-engineering·DomainSoftware Engineering
covers2
- skill-area:multi-app-orchestration·SkillAreaMulti-App Orchestration
- skill-area:multi-turn-tool-use·SkillAreaMulti-Turn Tool Use
Incoming edges
belongs_to_benchmark1
- test-set:appworld-750-cross-app-tasks·TestSetAppWorld 750 cross-app tasks