Agents' Last Exam (ALE) Benchmark

What is Agents' Last Exam (ALE)?

A living benchmark of 1K+ long-horizon professional tasks with verifiable outcomes, covering 55 sub-fields in 13 industry clusters based on O*NET / SOC 2018.

Released

2026-06-03

Evaluates

General AI, Language & Knowledge, cs.AI

Openness

unknown

Importer

Claire Radar

Review status

ai-reviewed

Reported scores

0

Source provenance

• Original evidence https://arxiv.org/abs/2606.05405

Agents' Last Exam (ALE) paper and code

• Paper https://arxiv.org/abs/2606.05405
• Code https://github.com/rdi-berkeley/agents-last-exam
• Website https://agents-last-exam.org

Which sources cite Agents' Last Exam (ALE)?

• Agents' Last Exam (ALE)

Related General AI, Language & Knowledge, cs.AI benchmarks

• $\tau^\tau$-Bench: An Environment for End-To-End, Realistic Agent Construction
• 3DCodeBench
• 5-Dialects-BN
• A Large-Scale Dataset and Benchmark
• A3S-Bench
• AARRI-Bench
• ABC-Bench
• ActBench