News
Astra for Law Achieves Top Rankings in Legal Intelligence Benchmarks
Astra for Law, powered by GPT-6.2 Astra, leads legal AI benchmarks with 95% fewer hallucinations and top scores in Harvey LAB-AA and Vals AI Legal Research Bench.
Astra for Law, a specialized configuration of OpenAI's GPT-6.2 Astra model, has achieved leading rankings in multiple legal intelligence benchmarks. The model, tailored for professional legal work, earned top marks in the Harvey LAB-AA, LAB: USCL, and Vals AI Legal Research Bench evaluations. OpenAI announced on November 17, 2026, that all ChatGPT Enterprise customers can now request access to Astra for Law through the Trusted Access Program, including API integration.
Performance Highlights
In the Harvey LAB-AA benchmark, which evaluates a model's ability to complete legal assignments across 24 areas of law, Astra for Law ranked first, earning full credit on 20.8% of assignments. This performance was 59% higher than the top non-OpenAI competitor, Muse Wonder Full, which achieved a 13.1% success rate. Astra for Law also demonstrated 95% fewer hallucinations per assignment compared to Claude Opus 5.5, a competing model.
On the Vals AI Legal Research Bench, Astra for Law improved its research accuracy significantly, with the share of answers passing all checks rising from 46% to 53.8% in under two months. These advancements were supported by a 33% expansion of the model's legal search index since its initial introduction in September 2026.
Technical Advancements
Astra for Law combines GPT-6.2 Astra with a dedicated legal search index and custom instructions for legal analysis and writing. The model is designed to assist lawyers in researching legal questions, developing arguments, and drafting documents. Its legal search index includes over 230 million URLs, covering U.S. case law, statutes, regulations, court rules, and administrative decisions, with daily updates to ensure comprehensive coverage.
Compared to its predecessor, GPT-6.2 Astra, Astra for Law earned full credit on LAB-AA tests nearly 40% more often, highlighting its enhanced legal reasoning capabilities. Additionally, the model achieved these results with 54% less output per assignment than the highest-scoring non-OpenAI competitor, demonstrating efficiency alongside accuracy.
Strategic Importance
OpenAI emphasized the importance of advancing legal intelligence to support the critical role lawyers play in shaping commerce, defending rights, and drafting laws. By prioritizing legal research and development, Astra for Law aims to equip legal professionals with state-of-the-art tools to navigate the complexities of modern legal systems.