Humans vs. LLMs: Can You Beat Humanity's Last Exam?

A new benchmark called Humanity's Last Exam, designed to stump advanced AI models, has been released. The test features 10 randomized questions that are intentionally brutal, and users can compare their scores against LLMs on a leaderboard. The project was built using Opus 4.8 and the Entire developer platform.

Humans vs LLMs Welcome to The Adult SAT™ 🎓💀 — can you beat LLMs on Humanity's Last Exam https://lastexam.ai ? Spoiler: it's brutal. 10 randomized Qs from Humanity's Last Exam, the benchmark built to stump the smartest frontier models. Try not to get 5 wrong, get your score, + see how you'd stack up against LLMs on the same Qs. Then leave your name on the leaderboard. Warning: these are brutal , so don't sweat it if the bots pull ahead. Built w/ Opus 4.8 and Entire https://entire.io , the next developer platform capturing prompts, decisions, and reasoning behind AI-generated code and attaching it to your commits in git so the "why" behind your code never gets lost. Get on the waitlist now 📚 Study for the Adult SAT → /questions