Benchmark RadarTravelEval
TravelEval evaluates LLM-powered travel planning agents in a realistic sandbox with accommodation pricing and intercity transport data, using six dimensions: accuracy, compliance, temporality, spatiality, economy, and utility. It simulates complete plans with API-integrated geographic information and queuing time.
- Released
- 2026-05-31
- Readiness
- Paper only
- Primary field
- General AI
Why it matters
Existing travel planning benchmarks overemphasize constraint compliance, lack real-world data coverage, and miss global plan quality. TravelEval offers a multi-dimensional framework to compare agent planning capabilities, aiding model selection for complex travel tasks.
Motivation
The development of Large Language Models (LLMs) has significantly improved travel planning applications, yet evaluating such models is limited by existing benchmarks' limitations: 1) overemphasis on constraint compliance, neglecting multi-dimensional qualities like spatio-temporal cost; 2) datasets lacking real-world authenticity and coverage in key areas (e.g., lodging, transport); and 3) isolated daily plan assess…
Primary resources
Benchmark Radar records only publicly supported details and links back to primary sources for verification.