Artificial Analysis (@ArtificialAnlys)T1·68 pts
- Search is a critical tool for agents, with projected agent search volume exceeding human usage by year-end
- The index evaluates performance using 1700 tasks and three specific benchmarks: BrowseComp, DeepSearchQA, and AA-Omniscience
- Exa (auto) and Perplexity (medium) currently lead in performance with top scores of 80