OpenAI launches Astra for Law, built on GPT-6 Astra

OpenAI introduced Astra for Law, a foundation that law firms and legal technology companies can build products and workflows on top of. It combines GPT-6 Astra, OpenAI's most powerful model, with settings, tools and context tailored to professional legal work. API customers including Harvey and Legora will be able to build on it directly.

The core addition is a legal search index that can search U.S. case law, statutes, regulations, court rules and administrative decisions across a corpus of more than 230 million URLs, updated daily. Through a partnership with Free Law Project, the nonprofit behind CourtListener, the index incorporates a case-law collection covering more than 99.9% of published U.S. precedential case law. OpenAI says the index complements, rather than replaces, licensed research products such as those from Thomson Reuters.

To measure the effect, OpenAI tested the complete Astra for Law setup on 200 U.S. legal research questions drawn from the private validation set of Vals AI's Legal Research Bench. At the highest reasoning effort for both systems, Astra for Law passed the benchmark's overall correctness check on 54.0% of questions, against 38.7% for GPT-6 Astra using web search alone, a 40% relative improvement. On case-law-focused questions, Astra for Law found 24% more reference cases than the web-search baseline, and on an audited set of target passages it retrieved up to 54% more relevant passages from the correct court opinions, at the same reasoning effort.

Beyond search, Astra for Law ships with custom instructions for legal analysis and writing, meant to help apply research to a client's facts: developing arguments or deal terms, distinguishing a court's holding from other observations, or explaining how a contract exception shifts risk between parties.

Access starts narrow. Astra for Law will initially go to selected law firms through a Trusted Access Program in ChatGPT and Codex, with API access coming soon; it will appear in the ChatGPT model picker as "GPT-6 Astra Law" and in the API as gpt-6-astra-law. Eligible firms in the program get Zero Data Retention on the API, and their ChatGPT Enterprise usage is excluded from human review by default. OpenAI is also working with Latham & Watkins on information permissions, ethical walls, client instructions and firm oversight, and with Wachtell, Lipton, Rosen & Katz to explore how AI can support more sophisticated legal judgment work.

Several firms have already built custom tools on ChatGPT Enterprise with OpenAI's forward-deployed engineers. Sullivan & Cromwell built an agreement analyzer that draws on the firm's negotiating playbooks and precedents to flag risks in a deal and draft redlines and client advice. Ropes & Gray built a deal diligence system that helps lawyers work through a data room, trace findings to source, and flag questions such as whether key customer contracts require notice or consent. Cooley built a tool called GO Public for capital-markets work, covering everything from drafting an IPO filing to tracking how changes ripple through it.

OpenAI also launched 26 ecosystem plugins connecting ChatGPT to specialist legal tools, including Relativity and Clio, plus iManage (drafting and filing negotiation briefs), Intapp (surfacing activity that may need a time entry) and DeepJudge (comparing prior deals); Thomson Reuters is bringing HighQ matter context into ChatGPT and previewing a forthcoming CoCounsel Legal connector. Alongside these, 9 community plugins built by lawyers and legal engineers at LegalQuants, LECG and Skills.law bring 47 custom skills that other practitioners can adapt. Separately, ChatGPT for Word is now generally available, letting lawyers proofread, get suggested edits and catch formatting issues inside the tool they already draft in.

OpenAI did not disclose pricing for Astra for Law, the Trusted Access Program or the plugins, and did not name the "other frontier models" it says Astra for Law beats at matching fact patterns to precedent.

Key facts

Astra for Law pairs GPT-6 Astra with a legal search index covering more than 230 million URLs of U.S. case law, statutes, regulations, court rules and administrative decisions, built with Free Law Project/CourtListener data covering over 99.9% of published U.S. precedential case law.
On 200 U.S. legal research questions from Vals AI's Legal Research Bench, Astra for Law passed the correctness check on 54.0% of questions versus 38.7% for GPT-6 Astra using web search alone, a 40% relative improvement; it also found 24% more reference cases and retrieved up to 54% more relevant passages.
It launches through a Trusted Access Program in ChatGPT and Codex for selected firms, with Zero Data Retention on the API and ChatGPT Enterprise usage excluded from human review by default; it will be gpt-6-astra-law in the API, with broader API access coming soon.
API customers Harvey and Legora will build on it directly, while Sullivan & Cromwell, Ropes & Gray and Cooley have already built custom tools on ChatGPT Enterprise, including an agreement analyzer, a deal diligence system and an IPO-prep tool called GO Public.
OpenAI launched 26 ecosystem plugins linking ChatGPT to tools like Relativity, Clio, iManage, Intapp, DeepJudge and Thomson Reuters, plus 9 community plugins with 47 custom skills, and made ChatGPT for Word generally available.

Why it matters

Astra for Law is OpenAI packaging its top model as a vertical product rather than just improving the general model: a legal search index, legal-analysis instructions and firm-specific controls are bundled together and offered through a controlled access program. It is a template for how OpenAI plans to sell frontier capability into regulated, high-stakes professions, with law firms as the first target.

Who it affects

Law firms and legal technology companies building directly on the platform, starting with API customers Harvey and Legora. Large firms are already named as design partners and early builders: Sullivan & Cromwell, Ropes & Gray, Cooley, Latham & Watkins, and Wachtell, Lipton, Rosen & Katz. Plugin partners include iManage, Intapp, DeepJudge and Thomson Reuters, alongside community developers at LegalQuants, LECG and Skills.law, and any lawyer who drafts in Microsoft Word.

How to use it

Access is initially limited to selected firms through the Trusted Access Program in ChatGPT and Codex; API access is coming soon. The model appears as "GPT-6 Astra Law" in the ChatGPT model picker and as gpt-6-astra-law in the API. Firms can extend it with 26 new ecosystem plugins connecting to tools like Relativity and Clio, or with 9 community plugins and 47 custom skills. No pricing has been disclosed for the model, the access program or the plugins.

How solid is it

The headline numbers come from OpenAI's own test: 200 U.S. legal research questions from the private validation set of Vals AI's Legal Research Bench, run at the highest reasoning effort for both systems. Astra for Law passed the correctness check on 54.0% of questions against 38.7% for GPT-6 Astra with web search alone, a 40% relative improvement, plus 24% more reference cases and up to 54% more relevant passages on separate, smaller audited subsets. OpenAI does not disclose the methodology behind those subset figures beyond the 200-question total, and the benchmark and comparison were run and reported by OpenAI itself.

Risks and caveats

No pricing is disclosed for Astra for Law, the Trusted Access Program or the plugins. OpenAI does not name the "other frontier models" it says Astra for Law outperforms at matching fact patterns to precedent, and gives no calendar date for the wider API rollout beyond "coming soon." The performance claims are OpenAI's own benchmark results rather than independent findings.

“The result is a stronger research foundation for advising on a deal, assessing a dispute, or developing a legal strategy, with reliable authorities the lawyer can examine for herself.”

— OpenAI