Harvey LAB-AA: Evaluating AI Agents on Real-World Legal Work
Harvey LAB-AA evaluates AI models on real-world legal tasks, providing critical insights for engineers.
Harvey LAB-AA evaluates the performance of AI models on real-world legal tasks. The top-performing model, Claude Fable 5, achieves a 14.2% all-pass rate, while many others struggle to meet task requirements fully. This highlights the importance for engineers to understand the potential and limitations of AI systems in legal processes.
This synthesis was produced from its source by AI; there is no human editor or manual review step. How we work