OpenAI launches Astra for Law with a 230 million-URL search index

Why it matters

OpenAI is moving beyond selling model access and into the legal workflow layer. Astra for Law gives Harvey, Legora and other partners ChatGPT distribution while placing OpenAI's search and configuration underneath their products.

OpenAI launched Astra for Law on September 17th, packaging its flagship GPT-6 Astra model with legal instructions, research infrastructure and integrations built for law firms and legal technology vendors.

https://x.com/OpenAI/status/2100679992720142459

OpenAI said in a five-post thread on X that the offering combines GPT-6 Astra with instructions for legal analysis and writing, settings intended for longer and more thorough work, and a new Legal Search Index. OpenAI says the index searches U.S. case law, statutes, regulations, court rules and administrative decisions across more than 230 million URLs.

Astra for Law will initially be available to selected firms through a Trusted Access program in ChatGPT and Codex. OpenAI says API access will follow, without giving a launch date.

The release moves OpenAI deeper into the application layer of legal AI. Astra for Law offers firms a maintained legal configuration instead of leaving each customer or software vendor to assemble prompts, retrieval systems and model settings around the underlying model. OpenAI is betting that a common foundation can become the default substrate for legal workflows, while firms preserve their own procedures and institutional knowledge in the applications built above it.

Law firms helped build the workflows

OpenAI named three systems developed with large U.S. law firms. Sullivan & Cromwell built an agreement analyzer with OpenAI engineers. Ropes & Gray developed an M&A diligence system. Cooley built GO Public, a workflow for preparing companies for initial public offerings.

Those examples show where OpenAI expects adoption to begin: document-heavy, repeatable assignments where firms can encode their own standards while lawyers retain responsibility for the output. OpenAI described Astra for Law as supporting lawyers' expertise and judgment, stopping short of presenting the system as a substitute for either.

OpenAI also launched 26 partner-built plugins and 47 community plugins for legal work in ChatGPT. The named partners include Thomson Reuters (@thomsonreuters), Harvey (@harvey), Legora (@WeAreLegora) and iManage (@imanageinc).

The partner list captures the tension in OpenAI's strategy. Harvey and Legora already sell specialist legal AI products, while Thomson Reuters and iManage control valuable legal content and document workflows. OpenAI is giving those vendors a distribution channel inside ChatGPT while building its own legal search, model configuration and workflow layer alongside them.

That arrangement lets OpenAI occupy the platform position. Partners can supply specialist knowledge, firm integrations and finished applications; OpenAI supplies the model, interface and maintained legal configuration. If the model layer becomes standardized, legal technology vendors will have to distinguish themselves through proprietary content, access controls, audit trails and workflow design rather than basic model access.

The search index carries the harder test

GPT-6 Astra launched on September 3rd as OpenAI's model for demanding professional and computer-based work. OpenAI's API documentation lists a 1.05 million-token context window, tool support including web and file search, and reasoning settings ranging from low to max.

Astra for Law narrows those general capabilities around a profession where research quality is measured by authority, currency and traceability. The figure of 230 million URLs describes the size of the search surface. It does not establish how the index handles duplicate documents, subsequent legal history, conflicting authorities or differences between controlling and persuasive precedent.

Those details will determine whether Astra for Law becomes a dependable research layer or another drafting assistant that requires lawyers to independently reconstruct its work. Legal users need to know which material was searched, why an authority was selected and whether a cited rule remains valid. A larger corpus helps only when the system can retrieve the right authority and show enough of its work for a lawyer to verify it.

OpenAI's controlled rollout gives selected firms an early role in shaping those mechanics. It also keeps the initial deployment inside environments where OpenAI and participating firms can set permissions and monitor how lawyers use the system before API access allows legal technology developers to build Astra for Law into their own products.

The product arrives two weeks after the broader Astra release and turns OpenAI's flagship model into a vertical package with its own search layer, partner channel and maintained configuration. That puts OpenAI closer to the legal workflows its specialist partners sell, even as OpenAI asks those same partners to build inside ChatGPT.