OpenAI has released Astra for Law, a specialized version of its GPT-6 Astra model designed specifically for legal research and document drafting. The new tool isn't a completely new model, but rather GPT-6 Astra wrapped with a legal search index covering more than 230 million URLs of U.S. case law, statutes, regulations, and court decisions, with sources added daily.

How Does Astra for Law Perform Compared to Standard AI Search?

OpenAI tested Astra for Law against its standard GPT-6 Astra model with web search capabilities, using 200 questions from a private validation set created by Vals AI's Legal Research Bench. The results show a meaningful gap in performance. Astra for Law passed the overall correctness check on 54% of questions, compared to 38.7% for GPT-6 Astra with web search alone, with both systems running at their highest reasoning effort. On case-law questions specifically, Astra for Law found 24% more relevant reference cases. In a separate audited evaluation, it retrieved up to 54% more target passages from the correct legal opinions.

These improvements matter because legal research requires precision. A lawyer searching for precedent needs not just any relevant case, but the right case with the exact passage that supports their argument. Missing a critical precedent or misreading a statute can have serious consequences for clients.

What Capabilities Are Built Into the New Legal AI Tool?

Astra for Law includes several features designed specifically for how lawyers work:

• Legal Research: Searches across U.S. case law, statutes, regulations, court rules, and administrative decisions, with the Free Law Project's CourtListener database supplying case law data.
• Document Analysis: Reads and reviews contracts, briefs, and agreements to quickly surface key terms and related precedents.
• Matter Management: Organizes documents within ChatGPT Enterprise and saves new or updated work back to legal platforms while maintaining version history.
• Contextual Intelligence: Uses what OpenAI calls the "Context Fabric" to reason over connected organizational knowledge, surfacing related work and prior matters rather than isolated documents.

Initial access to Astra for Law is limited through a Trusted Access program in ChatGPT and Codex, where it appears in the model picker as GPT-6 Astra Law. An application programming interface (API) version, gpt-6-astra-law, is planned for later release, with legal AI companies Harvey Technologies and Legora named as initial API customers.

How Are Law Firms Integrating ChatGPT Into Their Workflows?

The launch includes 26 partner-built plugins connecting ChatGPT to legal tools like Relativity, Clio, iManage, and DeepJudge. Thomson Reuters is bringing its HighQ matter context into ChatGPT and previewing a connector for its CoCounsel Legal product. Additionally, nine plugins built by lawyers and legal engineers outside the vendor ecosystem launched with 47 custom skills.

Major law firms are already building custom tools on ChatGPT Enterprise. Sullivan & Cromwell developed an agreement analyzer that pulls its negotiating playbooks and selected precedents into contract reviews and generates proposed redlines. Ropes & Gray built tools for deal diligence, while Cooley created GO Public, a tool for initial public offering preparation including drafting.

"As AI becomes more open and interoperable, the value is not in connectivity alone. Legal professionals need more than access to information. They need trusted intelligence, relevant enterprise and matter context, purpose built legal capabilities, and the governance required for high stakes work," said Joel Hron, Chief Technology Officer at Thomson Reuters.

Joel Hron, Chief Technology Officer at Thomson Reuters

What Role Does Data Governance Play in Enterprise Legal AI?

One critical concern for law firms is data security and client confidentiality. For eligible firms, Astra for Law carries zero data retention on the API, meaning OpenAI doesn't store queries or results for training purposes. ChatGPT Enterprise usage is excluded from human review by default. Latham & Watkins is working directly with OpenAI on governance design for information permissions, ethical walls, and client instructions.

This governance focus extends to iManage's new integration with ChatGPT Enterprise. iManage announced a plugin that gives knowledge workers secure access to governed iManage content from within ChatGPT Enterprise while maintaining existing permissions, ethical walls, and audit trails. The plugin allows users to search authorized content, read and review documents, organize materials, and save new work back to iManage while keeping the original and full version history intact.

"Our customers want the freedom to use the AI tools that best fit how their organizations work, without compromising the governance around their most important knowledge. This integration with ChatGPT makes it even easier to bring governed iManage knowledge into one of the AI environments they choose to use, while reflecting our broader open ecosystem strategy of giving customers choice and flexibility," said Neil Araujo, Chief Executive Officer of iManage.

Neil Araujo, Chief Executive Officer of iManage

OpenAI's Jason Boehmig, General Manager of Legal Industry, framed the partnership differently, emphasizing that the future of legal AI depends on what makes each firm unique. "We believe the next generation of AI for law will be built around what makes each firm distinctive, its expertise, judgment, and way of working. By building alongside iManage, we're turning that expertise into AI lawyers can use every day, creating new ways for firms to serve their clients and shape the future of legal work," he stated.

Jason Boehmig

OpenAI described this launch as the beginning of a long-term investment in the legal sector. GPT-6 Astra itself began rolling out two weeks prior, on September 3, 2026. The combination of improved legal research performance, firm-specific customization, and governance controls suggests that AI is moving beyond general-purpose chatbots into specialized tools designed for high-stakes professional work where accuracy and accountability matter most.