List your company

Task-oriented Plan Retrieval on Average across domains Single-turn

83Pass RateSemantic XPath
27.8842.1956.570.81Mar 1, 2026

Evaluation results

MethodLinks
Semantic XPathModel=GPT-5 mini, Scorer=Entail2026.03835,691
In-contextModel=Gemini 3 Flash, Scorer=–2026.037925,387
Semantic XPathModel=Gemini 3 Flash, Scorer=Entail2026.0370.75,830
In-contextModel=GPT-5 mini, Scorer=–2026.037026,216
Semantic XPathModel=GPT-5 mini, Scorer=Sem2026.0366.36,166
Semantic XPathModel=Gemini 3 Flash, Scorer=Sem2026.0364.36,043
Flat RAGModel=Gemini 3 Flash, Scorer=–2026.0337.73,361
Flat RAGModel=GPT-5 mini, Scorer=–2026.03303,250