OpenAI’s Astra for Law is a legal configuration of GPT-6 Astra, not a new base model

OpenAI launched Astra for Law, a legal-work product that combines the GPT-6 Astra model with a dedicated legal search index and legal-specific instructions rather than a separately trained base model. OpenAI says the configuration scored 54% on the overall correctness check of a 200-question legal research benchmark from Vals AI's Legal Research Bench, versus 38.7% for GPT-6 Astra using web search alone, and searches more than 230 million URLs covering U.S. cases, statutes, regulations, court rules and administrative decisions. Access is limited to selected U.S. firms through Trusted Access in ChatGPT and Codex, with the API planned but no published launch date or pricing, and OpenAI says eligible firms receive zero data retention on the API while it works with Latham & Watkins on permissions and ethical walls.

OpenAI’s Astra for Law is a legal configuration of GPT-6 Astra, not a new base model - Astra for Law combines GPT-6 Astra with a legal search index and legal-specific instructions; it is a product configuration and workflow foundation, not a separately trained base model. <sup \ 1\ https://openai.com/index/astra-for-law/ </sup <sup \ 2\ https://openai.com/index/astra-for-law/ </sup - The system searches more than 230 million URLs covering U.S. cases, statutes, regulations, court rules and administrative decisions. OpenAI says it scored 54% on the overall correctness check in a 200-question legal research benchmark, versus 38.7% for GPT-6 Astra using web search alone. <sup \ 2\ https://openai.com/index/astra-for-law/ </sup - Access is limited to selected U.S. firms through Trusted Access in ChatGPT and Codex. The API is planned but has no published launch date or pricing. <sup \ 3\ https://help.openai.com/en-us/articles/20001528-astra-for-law </sup - OpenAI says eligible firms receive zero data retention for API use, while ChatGPT Enterprise usage is excluded from human review by default. It is also working on permissions and ethical walls with Latham & Watkins. <sup \ 4\ https://openai.com/index/astra-for-law/ </sup OpenAI has launched Astra for Law, a legal-work product built by combining its GPT-6 Astra model with a dedicated legal search index, legal-specific instructions and controls for confidential client work. The company is positioning it as infrastructure that law firms and legal-technology companies can adapt, rather than as a standalone legal practice-management application. \ 1\ https://openai.com/index/astra-for-law/ The distinction matters. OpenAI calls Astra for Law a “new foundation,” but its documentation describes the system as GPT-6 Astra tailored for professional legal work. Independent coverage likewise characterizes it as GPT-6 Astra wrapped in legal search and workflow features, not a new base model. In ChatGPT, it appears as “GPT-6 Astra Law”; the planned API model name is gpt-6-astra-law . \ 2\ https://openai.com/index/astra-for-law/ \ 3\ https://help.openai.com/en-us/articles/20001528-astra-for-law A legal index around GPT-6 Astra Astra for Law can search U.S. case law, statutes, regulations, court rules and administrative decisions across a corpus of more than 230 million URLs, with new sources added daily. OpenAI says its work with the Free Law Project brings a CourtListener case-law collection covering more than 99.9% of published U.S. precedential case law into the research experience. \ 2\ https://openai.com/index/astra-for-law/ OpenAI tested the complete 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, the company says Astra for Law passed the benchmark’s overall correctness check on 54% of questions, compared with 38.7% for GPT-6 Astra using web search alone. It also found 24% more reference cases on case-law questions and retrieved up to 54% more relevant passages from correct opinions in an audited set. \ 2\ https://openai.com/index/astra-for-law/ Those results are company-reported and compare Astra for Law with another OpenAI system, rather than with Westlaw, Lexis+ or another commercial legal platform. They show the reported effect of adding a legal index and tailored instructions, but do not establish that Astra for Law is more accurate than incumbent tools. \ 2\ https://openai.com/index/astra-for-law/ \ 5\ https://openai.com/index/astra-for-law/ Controls, permissions and early deployments OpenAI says eligible firms in its Trusted Access program receive zero data retention on the API, while ChatGPT Enterprise usage is excluded from human review by default. The company is working with Latham & Watkins on information permissions, ethical walls, client instructions and firm oversight. \ 4\ https://openai.com/index/astra-for-law/ The announcement does not spell out a complete permission matrix, matter-level access policy or general-availability timetable. OpenAI’s Help Center says the initial offering is for selected U.S. law firms and eligible lawyers or people working under their supervision. \ 3\ https://help.openai.com/en-us/articles/20001528-astra-for-law OpenAI also describes firm-specific applications built with Sullivan & Cromwell, Ropes & Gray and Cooley. Sullivan & Cromwell has tested an agreement analyzer using negotiating playbooks and selected precedents; Ropes & Gray built a diligence workflow; and Cooley built GO Public for IPO preparation. OpenAI separately identifies Harvey and Legora as API customers that will be able to build on Astra for Law. \ 1\ https://openai.com/index/astra-for-law/ The launch includes 26 partner-built plugins, including connections involving iManage, Intapp, DeepJudge and Thomson Reuters’ HighQ. OpenAI also lists nine community plugins containing 47 custom skills from LegalQuants, LECG and Skills.law. These are integrations and partner-built tools, not evidence that those companies are Astra for Law customers unless OpenAI identifies them as such. \ 6\ https://openai.com/index/astra-for-law/ How it compares with incumbent legal AI Astra for Law’s clearest distinction is its position in the stack. OpenAI supplies the model, a broad legal search index and the ability to connect external tools, while firms and legal-technology companies build matter-specific workflows. OpenAI says its index is intended to complement licensed specialist products from providers such as Thomson Reuters. \ 2\ https://openai.com/index/astra-for-law/ Incumbents increasingly combine proprietary legal content with workflow automation. Thomson Reuters’ CoCounsel Legal is grounded in Westlaw, Practical Law and an organization’s own documents, and offers research, drafting, document analysis and matter workflows in one environment. Its current product materials describe agentic research, Westlaw Brief Builder and document analysis of up to 10,000 files against 100 questions. \ 7\ https://legal.thomsonreuters.com/en/products/cocounsel-legal Lexis+ with Protégé similarly combines LexisNexis primary law, practical guidance, document analysis, Shepard’s citation validation and organization-specific documents. \ 8\ https://www.lexisnexis.com/en-us/products/lexis-plus-protege/legal-research.page The comparison suggests Astra for Law is initially more of a model-and-ecosystem offering than a replacement for those proprietary research platforms. That is an assessment of the products’ published positioning, not an independent performance comparison. OpenAI has not announced pricing, a general API release date or a broad self-serve plan. \ 2\ https://openai.com/index/astra-for-law/ \ 3\ https://help.openai.com/en-us/articles/20001528-astra-for-law Companies mentioned Further sources The stories that matter, in one email. Free — unsubscribe anytime.