OpenAI introduced Astra for Law on 17 September, a configuration of its GPT-6 Astra model with legal search, instructions for legal analysis and writing, and controls aimed at law firms. Selected firms receive it first through a Trusted Access programme in ChatGPT and Codex, with the API to follow as gpt-6-astra-law. The launch matters because OpenAI is not only selling a model to law firms but also building the legal research index underneath it, which puts it alongside the vendors whose content law firms already license.
What sits under the model
The configuration matters less than the index beneath it. OpenAI has built a legal search index covering United States case law, statutes, regulations, court rules and administrative decisions, spanning more than 230 million URLs, with sources added daily. Much of the case law comes from the Free Law Project, the nonprofit behind CourtListener; OpenAI says that collection covers more than 99.9% of published US precedential case law. The company frames the index as complementary to licensed content and specialist products firms rely on from providers such as Thomson Reuters, whose CTO Joel Hron supplied a supportive quote for the launch. Thomson Reuters owns Westlaw, sells CoCounsel and has spent this year building its own legal models.
OpenAI tested the configuration on 200 US legal research questions from the private validation set of Vals AI's Legal Research Bench. At the highest reasoning effort, Astra for Law passed the overall correctness check on 54.0% of questions, against 38.7% for GPT-6 Astra with web search alone, which OpenAI presents as a 40% relative improvement. On case-law questions the configured system found 24% more reference cases and retrieved up to 54% more relevant passages from the correct opinions. The testing is OpenAI's own, as is the side-by-side comparison it published against Claude Fable 5.1, in which OpenAI says the rival model cited a holding an appeal court had overturned and in a second example reported finding no relevant case at all. No independent party has audited either result.
Alongside the model, OpenAI launched 26 partner-built plugins connecting ChatGPT to tools firms already use, including Relativity, Clio, iManage, Intapp and DeepJudge, plus nine community plugins from LegalQuants, LECG and Skills. law with 47 custom skills. ChatGPT for Word became generally available the same day. Six vendors issued releases between 16:00 and 16:15 New York time, and five carried the same quote from Jason Boehmig, OpenAI's general manager for the legal industry, word for word. Each release also draws a line: time entry continues in Laurel, deeper workflows continue in Litera, Relativity keeps substantive analysis inside its own platform, and Consilio passes only material responsive to the question asked.
What this means for business
For law firms and corporate legal departments, the practical change is that legal research, drafting support and access to firm tools arrive through one surface, ChatGPT, rather than through separate subscriptions. A small firm without a research budget gets a configured model and a case-law index it could not assemble itself, while a large firm with existing Westlaw or CoCounsel contracts faces a choice about which subscription keeps the work. The governance terms matter as much as the capability: eligible firms in the Trusted Access programme get Zero Data Retention on the API, and OpenAI excludes ChatGPT Enterprise usage from human review by default. OpenAI is also working with Latham & Watkins on information permissions, ethical walls, client instructions and firm oversight.
What the launch does not settle is price, which firms are inside Trusted Access, or when any of this reaches Europe; the search index covers United States law only, and the announcement does not address the sharper governance questions European firms face. The benchmark result also cuts both ways: a 54.0% pass rate on the correctness check means the configured system fails on nearly half the questions, and the comparison against a rival model comes from OpenAI itself. A firm evaluating Astra for Law should ask how the index is updated and versioned, what happens to a citation that a later court overturns, how Zero Data Retention interacts with its own record-keeping rules, and whether the plugins it depends on will keep the underlying workflow or only the interface.
Harvey, valued at $15.6bn this month, will build on Astra for Law through the API, as will Legora, which has been chasing a $10bn valuation; Harvey's head of applied research, Niko Grupen, praised its citation precision in a statement OpenAI published. OpenAI's own framing is that building on it should mean getting more from the ecosystem, not replacing it. The sign to watch is whether the best-funded legal AI companies keep shipping their own destinations on top of OpenAI's infrastructure once the index, the model and the plugin directory sit with one vendor, or whether they begin to look like interfaces to someone else's research layer.
