AIToday
Large Language ModelsAI Business & IndustryOpenAI BlogPublished: Sep 18, 2026, 06:00 JST

OpenAI launches Astra for Law on GPT‑6 Astra

OpenAI launches Astra for Law on GPT‑6 Astra

3 Key Points

  1. What happened

    OpenAI introduced Astra for Law, built on GPT‑6 Astra with a legal search index covering over 230 million URLs. It passed Vals AI's Legal Research Bench on 54.0% of questions, versus 38.7% for GPT‑6 Astra with web search alone.

  2. Why it matters

    Law firms may see AI research tools shift from experimental to practical for real legal work.

  3. What to watch

    The 54.0% pass rate is on a private validation set of 200 U.S. legal research questions, not a comprehensive measure of real-world legal accuracy. Whether Astra for Law gains traction with law firms will likely hinge on how well it handles the complexity and nuance of actual legal practice.

WHO IT HITSLaw firms and legal technology companies using AI for research, drafting, and diligence will see a new option from OpenAI. In-house legal teams evaluating AI tools for confidential client work may find the new privacy and governance controls relevant.

Not sure about something? Ask the AI

Summaries like this, in your inbox every morning.

Context & Analysis

OpenAI's move into legal AI builds on its frontier model work and addresses a sector where accuracy and confidentiality are paramount. The launch includes not only the core model but also a legal search index that draws on public case law and licensed content from providers like Thomson Reuters. Partnerships with firms such as Sullivan & Cromwell, Ropes & Gray, and Cooley show that OpenAI is co-developing tools with practitioners, which may help ensure the product fits real legal workflows.

The introduction of Astra for Law also brings a set of controls designed for law firms, including the special Trusted Access Program and work with Latham & Watkins on governance features. These steps suggest OpenAI is aware that legal professionals need assurances about data handling before adopting AI for client matters. The 26 ecosystem plugins and community skills further indicate an approach that connects AI to the tools lawyers already use, rather than trying to replace them.

How much this changes legal practice will likely depend on whether Astra for Law can deliver reliable results across the wide range of legal tasks, from research to drafting and diligence. The initial benchmark results are promising, but the true test will be how firms integrate these tools into their workflows and whether the productivity gains materialize in day-to-day work. For now, the launch signals that OpenAI sees law as a long-term investment and is willing to build alongside the profession.

FAQ
What is Astra for Law built on?
Astra for Law is built on GPT‑6 Astra, OpenAI's latest and most powerful model. It combines GPT‑6 Astra with a legal search index and instructions for legal analysis and writing.
How well does Astra for Law perform on legal research?
On 200 U.S. legal research questions from the private validation set of Vals AI's Legal Research Bench, Astra for Law passed the evaluation's overall correctness check on 54.0% of questions, compared with 38.7% for GPT‑6 Astra using web search alone.
Who can use Astra for Law?
Astra for Law will be initially offered to selected law firms through Trusted Access in ChatGPT and Codex. It will be coming soon to the API for legal technology companies like Harvey and Legora to build on.

Also reported by SiliconANGLE AI

Get the latest Large Language Models news every morning

For example, today's edition would include:

  • Google opens CC to families of six as shared AI agentSiliconANGLE AI · 2h ago
  • Anthropic shares pacing metrics, cites 30,000 agentsSiliconANGLE AI · 2h ago
  • Shiseido Japan uses AI to cut ingredient discovery time by 95%ITmedia AI+ · 2h ago

AI-summarized, only the topics you pick — one digest a day via Email, Slack, or Discord.

Free · 30 seconds with Google · unsubscribe anytimeWhat is AIToday? →

Ask AI

Ask AI anything about this article. Q&As are published on this page for other readers too.

Related Articles

Next articleWorkiva Amplify: traceable AI agents needed for finance