AI Benchmarks

The rigorous evaluations and datasets used to benchmark models.

AI Leaderboard Intelligence Indexv4.3.2

Incorporates evaluations: AIL-Briefcase v1.1, GDPval-AIL v2.1, AutomationBench-AIL, Terminal-Bench v4.0, SciCode, Humanity's Last Exam, GDP.pdf, CritPt, AIL-Omniscience, AIL-LCR v1.1

Terminal-Benchv4.0

Tests whether an agent can complete complex work through the terminal.

AutomationBench-AILv1.0

Zapier's business workflow automation benchmark.