OpenAI launches Astra for Law, a legal version of the GPT-6 Astra model
OpenAI launched Astra for Law, a legal version of the GPT-6 Astra model with an index of over 230 million URLs covering US case law and regulations. In the Legal Research Bench test, it achieved a 54 % success rate compared with 38.7 % for the standard GPT-6 Astra with web search.
OpenAI introduced Astra for Law, a version of the GPT-6 Astra model designed for legal practice. The product combines the model with a legal search index and instructions for legal analysis and writing.
The index covers US case law, statutes and regulations across more than 230 million URLs. The data comes from Free Law Project, which, according to its own claim, covers over 99.9 % of published US precedents. In a test conducted by OpenAI itself using Legal Research Bench from Vals AI, Astra for Law correctly answered 54 % of 200 questions, while the standard GPT-6 Astra model with ordinary web search achieved 38.7 %.
According to OpenAI, API customers such as Harvey and Legora can build on the product. Law firms have access to the Trusted Access program with privacy protections that include zero data retention. OpenAI is also launching 26 plugins for tools such as Relativity and Clio.
Why it matters
For lawyers and legal teams, this is a tool that, according to OpenAI, is intended to make case law searches and the preparation of supporting materials faster and more accurate – in the test, the company measured an improvement in accuracy compared with a standard model with web search, but an accuracy of 54 % means that outputs still need to be verified. For law firms, the availability of the Trusted Access program with zero data retention matters because it addresses the confidentiality of client information when deploying AI tools, and integration with existing software through 26 plugins makes it easier to incorporate the tool into existing workflows.
What was added since the original report
Verified updates
-
Astra for Law is built on the GPT-6 Astra model; The legal index covers 230+ million URLs of US case law and legal rules sourced from Free Law Project; Achieves a 54% success rate in Legal Research Bench, compared with 38.7% for GPT-6 without the index; The API is available to clients Harvey and Legora through the Trusted Access program with zero data retention; OpenAI launched 26 plugins for legal software such as Relativity and Clio
- Astra for Law is built on the GPT-6 Astra model
- The legal index covers 230+ million URLs of US case law and legal rules sourced from Free Law Project
- Achieves a 54% success rate in Legal Research Bench, compared with 38.7% for GPT-6 without the index
- The API is available to clients Harvey and Legora through the Trusted Access program with zero data retention
- OpenAI launched 26 plugins for legal software such as Relativity and Clio
Two audiences, two different impacts
What this means
For individuals
For lawyers and people conducting legal research, this is a tool that, according to the cited test, improves the accuracy of case law searches compared with a standard model with web search, but its outputs still need to be verified.
For a business
Law firms gain the option to deploy a specialized AI tool for research through the API (e.g. partners Harvey and Legora) or the Trusted Access program with zero data retention, addressing the confidentiality of client information when introducing AI into legal practice.
ProductivityCheck the original
Event sources
independently confirmed · 2 publishers, 1 independent. We count feeds from the same owner only once.