VECTOR WIREAI INTELLIGENCE
UTC
Refresh Models Deals Regulatory Sources

OpenAI bets GPT-6 Astra on legal market with 230M-URL search index

OpenAI's Astra for Law pairs GPT-6 Astra with a 230M-URL legal search index, scoring 54% on a legal benchmark — 15 points above the general model — and…

OpenAI on September 17 launched Astra for Law, a vertical AI platform that pairs GPT-6 Astra with a dedicated legal search index spanning more than 230 million URLs, targeting select law firms as its initial customers1,3.

The product combines the GPT-6 Astra model with specialized instructions for legal analysis and writing. OpenAI describes it as "our most powerful model, configured into a new AI foundation for law". The legal search index incorporates the CourtListener case database, which covers over 99.9% of published U.S. case law.

OpenAI validated Astra for Law using a private test set from the Vals AI legal research benchmark, consisting of 200 U.S. legal questions. At the highest reasoning intensity, Astra for Law achieved an overall accuracy rate of 54.0%, compared with 38.7% for GPT-6 Astra relying only on web searches — a gap of more than 15 percentage points. On case-law-oriented questions specifically, Astra for Law found 24% more relevant cases, and in audited target paragraph sets it extracted up to 54% more relevant paragraphs from correct court opinions.

OpenAI emphasized that Astra for Law is not an "AI lawyer". The platform is designed to serve lawyers and legal software companies on specific tasks: helping distinguish core reasoning from incidental comments in court decisions and actively surfacing unfavorable precedents.

The initial rollout targets select law firms, and OpenAI is setting its sights on the 200 largest U.S. law firms as a broader market2. The product includes a dedicated legal sales channel.

ANALYSIS A 54% accuracy rate on a 200-question legal benchmark, while a 15-point improvement over the general-purpose model, still means nearly half of answers are incorrect — a constraint that reinforces OpenAI's own framing of the tool as an aid to lawyers rather than a replacement. The launch arrives one day after OpenAI disclosed six model misalignment incidents and a new reporting framework[1], placing the company's first major vertical product debut alongside heightened scrutiny of its safety practices.