OpenAI Launches Astra for Law With a 230-Million-URL Legal Search Index
OpenAI launched Astra for Law, a specialized legal AI offering combining GPT-6 Astra with a purpose-built legal search index, aimed at law firms and legal technology companies building their own workflows on top of it.
Details
- Legal search index: covers more than 230 million URLs of U.S. case law, statutes, regulations, and administrative decisions, incorporating Free Law Project’s CourtListener data for over 99.9% of published U.S. precedential case law
- Benchmark performance: scores 54.0% on Vals AI’s Legal Research Bench versus 38.7% for standard GPT-6 with web search — a 40% relative improvement — and identifies 24% more reference cases on case-law questions
- Legal-specific capabilities: custom instructions support applying research to client facts, developing arguments, identifying weaknesses, and addressing precedent nuances
- Design partners: developed with Sullivan & Cromwell, Ropes & Gray, and Cooley, with governance input from Latham & Watkins and Wachtell, Lipton, Rosen & Katz
- Ecosystem: 26 plugins connecting to legal tools including Relativity, Clio, and Thomson Reuters
- Availability: initially through Trusted Access in ChatGPT and Codex for selected firms, with API access coming soon as “gpt-6-astra-law”
What happened next
Astra for Law extends OpenAI’s vertical-product strategy — following ChatGPT for Financial Services and its healthcare offerings — into legal research, betting that a purpose-built search index over near-complete case law coverage, rather than the base model alone, is what separates a usable legal research tool from a generic assistant.