The Pulse
OpenAI Launches Astra for Law With 230 Million URLs
OpenAI introduced Astra for Law, a GPT-6 Astra configuration built for law firms and legal technology companies. The product includes a legal search index covering more than 230 million URLs, 26 partner plugins and new controls for confiden
AI.info Team ·
OpenAI puts its legal index against established research systems
OpenAI is introducing Astra for Law as both a specialized model configuration and a challenge to the way legal software companies assemble research tools. The product uses GPT-6 Astra with legal instructions, a dedicated search index and integrations aimed at firms that already depend on specialist databases and matter-management systems.
The company says its new index is designed to complement licensed services such as Thomson Reuters rather than replace them. It searches U.S. case law, statutes, regulations, court rules and administrative decisions across more than 230 million URLs, with sources added daily. OpenAI says its work with the nonprofit Free Law Project brings CourtListener’s collection of more than 99.9% of published U.S. precedential case law into the research experience.
That positioning reflects the product’s intended buyer. Astra for Law is not launching as a general-purpose consumer chatbot with a legal skin; OpenAI describes it as a foundation for law firms and legal technology companies to build around their own expertise, workflows and review standards.
A benchmark gap separates Astra for Law from ordinary web search
OpenAI tested the full configuration on 200 U.S. legal research questions from the private validation set of Vals AI’s Legal Research Bench. At the highest reasoning setting, 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.
OpenAI characterizes the difference as a 40% relative improvement. On case-law questions, the legal configuration found 24% more reference cases than GPT-6 Astra with web search, while an audited set of target passages produced up to 54% more relevant passages from the correct court opinions.
The figures measure retrieval and answer quality on a private validation set, not the outcome of a court filing or a client matter. OpenAI says lawyers can examine the authorities behind the answers, while its legal instructions are intended to help the system distinguish a court’s holding from other observations, identify cases that weaken an argument and explain uncertainty.
Selected firms get access before the API
OpenAI will initially offer Astra for Law to selected firms through a Trusted Access program in ChatGPT and Codex. The model will appear in the model picker as “GPT-6 Astra Law” and is scheduled to come to the API as gpt-6-astra-law.
John Savva, a partner at Sullivan & Cromwell, said the firm previewed early versions of the system across litigation and transactional work. “Across both litigation and transactional matters, the models demonstrated impressive research depth and sensitivity to authority,” Savva said. “Even at this early stage, they felt like a significant step toward legal-focused AI that is carefully grounded in research that is both current and comprehensive.”
Harvey is also among the API customers that will be able to build on Astra for Law. Niko Grupen, Harvey’s head of applied research, said the company’s early testing showed strength in grounding answers in relevant authorities, citing them precisely and providing practical guidance.
Privacy controls are part of the product pitch
OpenAI is pairing the launch with controls aimed at firms handling confidential client information. Eligible firms in the Trusted Access program receive Zero Data Retention on the API, while ChatGPT Enterprise usage is excluded from human review by default, according to the company.
OpenAI is working with Latham & Watkins on information permissions, ethical walls, client instructions and firm oversight. Michael Rubin, chair of Latham’s AI Strategy Committee, said the collaboration is intended for environments that require governance and accountability as model capabilities increase.
The company is also working with firms to connect proprietary precedents and internal methods to custom applications. Sullivan & Cromwell built an agreement analyzer; Ropes & Gray built a deal-diligence system; and Cooley built GO Public, a tool for capital-markets work that helps carry changes across an IPO filing.
Twenty-six plugins extend Astra into existing legal systems
Astra for Law launches with 26 partner-built plugins connecting ChatGPT to tools used by legal teams. The integrations include Relativity, Clio, iManage, Intapp and DeepJudge. Thomson Reuters is bringing HighQ matter context into ChatGPT and previewing a future CoCounsel Legal connector.
OpenAI also lists nine community plugins from LegalQuants, LECG and Skills.law, along with 47 custom skills that practitioners can adapt. ChatGPT for Word is becoming generally available at the same time, adding proofreading, suggested edits and formatting checks inside Microsoft Word.
The arrangement gives OpenAI a role underneath products that firms already use while leaving room for vendors and law firms to build their own interfaces. OpenAI says firms can define permitted sources and review processes instead of adopting one fixed workflow.
OpenAI’s legal bet starts with research, not replacement
Astra for Law’s first release makes a specific claim: better retrieval and legal reasoning can improve the work around a matter without removing the lawyer from the decision. The product’s own examples focus on identifying authorities, tracing findings to source material, preparing redlines and carrying changes through transaction documents.
OpenAI is offering the technology first to selected firms and partners, with broader API access still ahead. Its immediate test will be whether the 230-million-URL index, firm-specific controls and connected legal tools produce answers that lawyers can verify and use within existing professional obligations.