General and professional work benchmark

Agents' Last Exam

A living benchmark built with 300+ industry experts that keeps adding tasks toward a 5,000 task target. Snorkel AI also hosts a leaderboard for it.

UC Berkeley RDI and RDI Foundation

At a glance

What the agent must do
The agent must complete long, economically valuable professional workflows on a computer, often in real industry software, with verifiable outcomes.
Size
1,500+ task corpus across 55 sub industries; about 150 public tasks
Scoring
Per task score from 0 to 1 against a hidden reference, reported as pass rate and score
How to run it
ale_run toolkit with an experiment YAML; sandboxes on Google Cloud VMs (recommended), AWS, QEMU/KVM or local Docker

Official sources

Official leaderboardBenchmark sitePapergithub.com/rdi-berkeley/agents-last-exam

Other general and professional work benchmarks

Harbor-IndexGAIAMETR Time HorizonsGDPvalAPEX-Agents

Run evals on your own agent

Benchmarks compare models and agents in general. To test your own agent, see the eval frameworks comparison.

Get the weekly agent stack update

New official MCP servers, spec changes and harness releases, checked against the source. One email a week, no fluff.

Reviewed Oct 9, 2026.