Epoch AI produces open datasets, benchmarks, and forecasts to explain AI’s progress, its underlying resources, and its economic impact through independent empirical research.
Products & media
Open datasets and interactive explorers track AI models, benchmarks, infrastructure, chip markets, companies, and adoption for researchers and policymakers.
A mathematics benchmark suite evaluates difficult problem solving through tiered questions, open research problems, and formally verified Erdős problems.
Commissioned research provides organizations with AI trend analysis, model evaluations, benchmark development, and technical briefings tailored to their questions.
A shared capability scale combines results across benchmarks to compare AI models and track performance improvements over time.
A coding benchmark developed with METR evaluates whether AI agents can recreate entire programs without seeing their source code.
A game-based benchmark measures how AI agents learn from repeated experience and written notes while playing Earthborne Rangers.
Cosigns
“I wouldn't have nearly as good an understanding of important trends in AI and their implications without Epoch.”
Team
Jobs
See all jobs for Epoch AI here.
