The Public Standard for
Real-World AI Benchmarks

A single measure of AI's potential economic impact — agentic model performance across finance, coding, legal, and tax tasks, weighted by each sector's share of U.S. GDP.

Can a model do the research that builds the next model? Autonomous LLM R&D scored against published, human, and frontier-model references.

Comparing native provider search against independent web-search tools on legal-research and finance-analysis tasks

A single measure of AI's ability to do science — agentic model performance on biology data investigation, mechanism discovery, and research workflows, weighted equally.

How model routers compare with single models on long agentic work in finance, coding, law and tax

Evaluating agents on legal research tasks across diverse areas of US law

Tests an agent's ability to complete legal work using documents, spreadsheets, presentations, and file-system tools.

Evaluating language models on a wide range of open source legal reasoning tasks.

Evaluating agents on core financial analyst tasks

Evaluating agents on Excel-based financial modeling tasks

Evaluating agents on research-grade US corporate tax questions

Can a decision model automate checking whether a 10-K excerpt supports, contradicts or says nothing about a claim?

Can models support the medical billing process?

Can models support doctors with their administrative work?

Can models respond safely as teens disclose concerns, resist advice, and ask for help?

Can models write math proofs that are formally verified?

Research workflow tasks contributed by practicing scientists across five domains

Can agents solve agentic biology data investigations?

Can agents rediscover sealed mathematical mechanisms through bounded experiments?

Can a model control a robot arm and a humanoid in a real-time physics simulation, from motor commands and camera images alone? Most tasks are held out.

Student Assessment with Generative Evaluation

How good are speech-to-text models at handling exact structured values in English workplace speech?

Can models build web applications from scratch?

Can models extend a working web application across future user requests?

Can language models reimplement working programs in another language?

Can language models rebuild programs from scratch?

Frontier-difficulty terminal tasks across software, science, ML, operations, hardware, security, and media

How important are skills for agents?

International Olympiad in Informatics

Do AI Systems Generalize to Visual Software Domains?

Can AI agents find security bugs and patch them?

Can AI agents work out what a real-world binary does without its source code?

Can an AI agent play six commercial video games with only a keyboard and a mouse? Half of them are held out.

Can an AI agent build and run a space program in Kerbal Space Program?

Which model can make the most money playing poker?

Can AI help people navigate SNAP benefits?