OpenAI packages Astra around U.S. legal research
OpenAI launched Astra for Law on 17 September as a specialised configuration of GPT-6 Astra for professional legal work. The product combines the base model with a dedicated legal search index and instructions for legal analysis and writing, rather than introducing a separate foundation model trained only for law.
The legal index covers U.S. case law, statutes, regulations, court rules and administrative decisions across more than 230 million URLs, with new sources added daily. OpenAI says its work with Free Law Project brings CourtListener coverage of more than 99.9% of published U.S. precedential case law into the research experience. The company also positions the index as complementary to licensed legal products rather than a replacement for them.
Access starts with selected U.S. law firms
Astra for Law is initially available to selected U.S. law firms through OpenAI's Trusted Access programme in ChatGPT and Codex. OpenAI says access is intended for eligible lawyers and people working under their supervision. In the model picker it appears as GPT-6 Astra Law.
OpenAI plans to expose the same configuration through the API as gpt-6-astra-law, but describes that access only as coming soon. It has not published an API launch date or pricing. For eligible firms, OpenAI says the offering includes Zero Data Retention on its API and that ChatGPT Enterprise usage is excluded from human review by default.
The launch adds legal plugins and firm-specific workflows
The product arrives with 26 partner-built plugins connecting ChatGPT to specialist legal tools, including systems from iManage, Intapp, DeepJudge, Relativity and Clio. OpenAI separately lists 9 community plugins from lawyers and legal engineers and 47 custom skills that practitioners can adapt for their own work. The official product page is the stronger source for those counts where launch-day social posts differ.
OpenAI is also working with selected firms on custom interfaces and integrations to proprietary data. The examples include agreement review, transaction diligence and public-offering preparation. Those firm-built workflows remain subject to the data permissions, review processes and professional controls chosen by each organisation.
OpenAI reports a benchmark gain, but the evaluation is private
OpenAI evaluated the full Astra for Law configuration on 200 U.S. legal research questions from a private validation set of Vals AI's Legal Research Bench. At the highest reasoning setting, OpenAI reports that Astra for Law passed the benchmark's overall correctness check on 54.0% of questions, compared with 38.7% for GPT-6 Astra using web search alone. It also reports finding 24% more reference cases on case-law questions and retrieving up to 54% more relevant passages from the correct opinions in an audited subset.
Those figures are OpenAI-run results on a private validation set, so they cannot yet serve as an independently reproducible measure of legal accuracy. OpenAI's own Help Center tells users to review answers and cited sources before relying on them. The launch establishes the product, search corpus, initial access model and integration surface; reliability across jurisdictions, practice areas and live firm workflows still needs external testing.