OpenAI is making a major push into the legal sector with the launch of Astra for Law, a specialized AI foundation designed for law firms and legal technology companies.
The new offering combines GPT-6 Astra, OpenAI’s most powerful model, with a legal search index and instructions tailored to legal analysis and writing. Rather than positioning Astra for Law as a standalone legal product, OpenAI is offering it as a foundation firms and technology providers can use to build their own applications and workflows.
Legal AI companies Harvey and Legora are among the API customers expected to build on Astra for Law. The platform will initially be available to selected law firms through a Trusted Access program in ChatGPT and Codex, with API access coming soon.
It will appear in OpenAI’s model picker as GPT-6 Astra Law and through the API as gpt-6-astra-law.
Legal Research Gets a Specialized Model
At the centre of Astra for Law is a new legal search index covering U.S. case law, statutes, regulations, court rules and administrative decisions across more than 230 million URLs.
The index incorporates the Free Law Project’s CourtListener collection, which covers more than 99.9 per cent of published U.S. precedential case law. OpenAI says it is designed to help the model locate relevant authorities, identify useful passages and assess how closely a decision applies to the matter being analyzed.
In testing on 200 questions from a private validation set of Vals AI’s Legal Research Bench, OpenAI said Astra for Law passed the benchmark’s overall correctness check on 54 per cent of questions. GPT-6 Astra using conventional web search passed 38.7 per cent, representing a 40 per cent relative improvement.
Astra for Law also found 24 per cent more reference cases on case-law questions and retrieved up to 54 per cent more relevant passages from the correct opinions, according to OpenAI.
The company says custom instructions then help the model apply that research to a client’s facts, develop arguments or deal terms and identify weaknesses or uncertainty. Potential tasks include distinguishing a court’s holding from other observations, addressing authorities that weaken an argument and examining how contractual exceptions redistribute risk.
The initial research index is focused on American law, an important limitation for Canadian firms evaluating the offering.
Building Around Firm Expertise
OpenAI is also working directly with major law firms to build applications around their internal precedents, methods and professional judgment.
Sullivan & Cromwell developed an agreement analyzer that uses the firm’s negotiating playbooks and selected precedents to identify risks, suggest redlines and prepare draft client advice. Ropes & Gray built a transactional diligence system, while Cooley created a tool to support companies through the IPO process.
The company is working with Latham & Watkins on governance features addressing permissions, ethical walls, client instructions and firm oversight. Eligible firms in the Trusted Access program will receive Zero Data Retention through the API, while ChatGPT Enterprise usage is excluded from human review by default.
OpenAI Connects With Legal Tech Platforms
OpenAI also unveiled 26 partner-built plugins connecting ChatGPT with established legal tools and information systems.
The ecosystem includes integrations with platforms such as Clio, Relativity, iManage, Intapp, DeepJudge and Thomson Reuters. Example workflows include saving a negotiation brief from ChatGPT into an iManage matter file, surfacing possible time entries through Intapp and comparing a proposed transaction with previous deals using DeepJudge.
Thomson Reuters is bringing HighQ matter context into ChatGPT and previewing a forthcoming CoCounsel Legal connector.
OpenAI also made ChatGPT for Word generally available, allowing legal professionals to proofread documents, receive suggested edits and identify formatting problems without leaving Microsoft Word.
The launch signals that OpenAI intends to become an underlying intelligence layer for legal work while leaving firms and legal technology companies to determine how that intelligence is deployed.
For legal AI providers such as Harvey and Legora, Astra for Law offers a more capable foundation on which to develop specialized products. For law firms, its appeal may rest on whether OpenAI can combine stronger legal reasoning with the confidentiality, governance and system integrations required for professional practice.