Astra for Law: OpenAI Builds AI for Legal Work

OpenAI has introduced Astra for Law, a specialized AI foundation designed for law firms and legal technology companies. The offering combines GPT-6 Astra with legal-specific instructions, settings, tools, and a new legal search index intended to help professionals conduct research and build AI workflows around legal expertise.

Announced on September 17, 2026, Astra for Law is initially being offered to selected law firms through OpenAI’s Trusted Access program in ChatGPT and Codex. API access is planned for a later stage, with OpenAI saying the system will appear as “GPT-6 Astra Law” in the model picker and as gpt-6-astra-law through the API.

Rather than introducing a general-purpose model alone, OpenAI is positioning Astra for Law as a configuration of its latest model specifically adapted for professional legal work.

Quick Summary

OpenAI introduced Astra for Law, a specialized AI foundation designed for legal professionals and legal technology companies.
It combines GPT-6 Astra with legal-specific instructions, reasoning settings and professional tools.
A new Legal Search Index searches U.S. case law, statutes, regulations, court rules and administrative decisions.
OpenAI says the index covers more than 230 million URLs, with sources added daily.
Law firms can build customized workflows for tasks such as agreement analysis, M&A diligence and IPO preparation.
The launch includes 26 partner-built plugins and 9 community plugins containing 47 custom skills.
Selected firms initially receive access through Trusted Access in ChatGPT and Codex.
API access is planned, allowing legal technology companies to build Astra for Law into their own products.

Astra for Law Adds a Legal Search Index

One of the main components of Astra for Law is a new legal search index designed to support legal research.

According to OpenAI, the index can search U.S. case law, statutes, regulations, court rules, and administrative decisions across a corpus of more than 230 million URLs, with sources added daily. OpenAI also says its work with Free Law Project brings a case-law collection covering more than 99.9% of published U.S. precedential case law into the research experience.

The search system is designed to help lawyers move from the facts of a matter to relevant legal authorities and supporting passages. That is particularly important for legal workflows in which identifying the correct authority is only part of the task; lawyers also need to evaluate relevance, understand the authority’s role, and examine the underlying material themselves.

OpenAI says the legal search index is intended to complement licensed content and specialist legal products used by firms, including services from providers such as Thomson Reuters.

OpenAI Reports Higher Legal Research Performance

OpenAI also published results from an internal evaluation of the complete Astra for Law configuration.

The company tested the system on 200 U.S. legal research questions from the private validation set of Vals AI’s Legal Research Bench. At the highest reasoning setting, OpenAI says Astra for Law passed the benchmark’s overall correctness evaluation on 54.0% of questions, compared with 38.7% for GPT-6 Astra using web search alone.

OpenAI describes that as a 40% relative improvement.

The company also reports that Astra for Law found 24% more reference cases on case-law-focused questions at the highest reasoning setting. On an audited set of target passages, OpenAI says the system retrieved up to 54% more relevant passages from the correct court opinions.

These figures are OpenAI-reported results from its stated validation setup, rather than an independent benchmark conducted by a third party. The results therefore provide context for how OpenAI says the legal configuration performs, but do not by themselves establish performance across the full range of real-world legal matters.

Legal Instructions Extend Beyond Search

Astra for Law is not limited to retrieving legal sources.

OpenAI says it combines GPT-6 Astra with custom instructions for legal analysis and writing. Those instructions are designed to help the model apply research to client-specific facts, develop arguments or deal terms, identify weaknesses and explain uncertainty.

The company gives examples including distinguishing a court’s holding from other observations, identifying cases that weaken an argument and analyzing how contractual exceptions can shift risk between parties.

This reflects a broader design approach in which the underlying model is combined with domain-specific instructions, information sources, and tools rather than treated as a standalone chatbot.

Law Firms Can Build Custom AI Workflows

OpenAI is also working directly with law firms to create specialized applications around their existing processes.

Sullivan & Cromwell has built an agreement analyzer designed to incorporate negotiating playbooks and selected precedents into contract review. The system can help identify risks that arise when provisions are considered together and generate proposed redlines and client advice for lawyers to review.

Ropes & Gray has developed a deal-diligence system centered on its approach to reviewing data rooms and identifying issues relevant to transactions.

Cooley has built GO Public, a system focused on capital-markets work and IPO preparation, including drafting support and identifying risks that may require management attention.

These examples illustrate how OpenAI is positioning Astra for Law as a foundation that firms can adapt to their own expertise and workflows rather than as a single standardized legal application.

Privacy and Governance Controls Target Legal Work

Because legal organizations routinely handle confidential client information, OpenAI is also emphasizing governance and data controls alongside model capabilities.

Through its Trusted Access Program for eligible law firms, OpenAI says Astra for Law includes Zero Data Retention on its API. At the same time, ChatGPT Enterprise usage is excluded from human review by default.

OpenAI is also working with Latham & Watkins on information permissions, ethical walls, client instructions, and firm oversight.

These controls are intended to address operational requirements that differ from those of ordinary consumer AI use, particularly when AI systems are incorporated into confidential legal workflows.

Astra for Law Connects to Legal Tools and Plugins

OpenAI is expanding the legal ecosystem around ChatGPT through partner-built and community-developed tools.

The launch includes 26 partner-built plugins connecting ChatGPT with specialist legal and business tools. OpenAI highlights integrations involving companies such as iManage, Intapp, DeepJudge, and Thomson Reuters.

For example, OpenAI says iManage can allow lawyers to draft a negotiation brief in ChatGPT and save it to a matter file. At the same time, Intapp can surface activities that may require time-entry review. Thomson Reuters is bringing HighQ matter context into ChatGPT and previewing a forthcoming CoCounsel Legal connector.

There is also a community layer. OpenAI says the launch includes nine community plugins from lawyers and legal engineers at LegalQuants, LECG, and Skills. law, containing 47 custom skills that practitioners can adapt or extend.

That distinction is important because the X posts describe 47 community plugins, whereas OpenAI’s official announcement specifies nine community plugins and 47 custom skills.

Availability and API Access?

Astra for Law is initially available to selected law firms through OpenAI’s Trusted Access program in ChatGPT and Codex.

In ChatGPT, the model will appear as GPT-6 Astra Law. OpenAI says the corresponding API model identifier will be gpt-6-astra-law, but API access is still listed as coming soon.

OpenAI says API customers, including Harvey and Legora, will be able to build on Astra for Law, allowing the legal capabilities to be incorporated into their own products and workflows.

The company also says it intends to maintain the legal configuration as its frontier models advance, allowing developers and legal technology companies to focus on their applications and workflows rather than rebuilding the legal layer themselves.

What Astra for Law Changes for Legal AI?

Astra for Law represents a shift from using a general-purpose AI model for legal tasks toward combining frontier model capabilities with domain-specific search, instructions, governance, and professional software.

Its initial deployment is deliberately limited to selected firms, while OpenAI develops the legal configuration alongside lawyers and legal technology partners. The company says it will continue evaluating the system and refining its model, settings, tools and instructions based on feedback from legal professionals.

For legal technology developers, the planned API availability could eventually provide another route for incorporating OpenAI’s legal configuration into specialized products. For law firms, the more immediate focus is on connecting AI with firm expertise, trusted sources, existing tools, and review processes.

Astra for Law therefore sits at the intersection of legal AI, AI-assisted research and workflow automation, with OpenAI treating domain-specific context and governance as important parts of the system rather than relying on the underlying model alone.

Also Read –

GPT-5.4 Mini by OpenAI: Faster AI Model for Coding and Agents

Codex App Automations: OpenAI Adds AI Workflow Automation

GPT-6 Astra: Features, Capabilities, Performance & What’s New?