CAVEAT Benchmark

What is CAVEAT?

Benchmark for computer-use agents in incentive-misaligned environments with nine marketplace environments and eight steering mechanisms.

Released

2026-09-23

Evaluates

General AI, Language & Knowledge, cs.AI

Openness

unknown

Importer

Claire Radar

Review status

ai-name-audit-deferred

Reported scores

0

Source provenance

• Original evidence https://arxiv.org/abs/2609.27273

CAVEAT paper

• Paper https://arxiv.org/abs/2609.27273

Which sources cite CAVEAT?

• CAVEAT

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

• $\tau^\tau$-Bench: An Environment for End-To-End, Realistic Agent Construction
• 2d-fet-bench
• 3DCodeBench
• 5-Dialects-BN
• A Large-Scale Dataset and Benchmark
• A Real-World CAD Segmentation Benchmark
• A3S-Bench
• AARRI-Bench