OpenAI has tailored GPT-6 into Astra for Law, a specialized model for professional legal work that combines a dedicated Legal Search Index with custom instructions for legal analysis and writing. The tool will initially roll out to selected U.S. law firms through Trusted Access and Codex, with API availability to follow, giving eligible lawyers and their supervised staff a new way to research case law, develop arguments, and stress-test deal terms.
The Legal Search Index draws on U.S. case law, statutes, regulations, court rules, and administrative decisions. It updates daily. Lawyers can use the model to surface cases that weaken a position or to trace how a contract exception shifts risk allocation. OpenAI said users should review answers and cited sources before relying on them.
How access and deployment work
Firms interested in early access must contact their OpenAI account team or OpenAI Sales. The model will appear as GPT-6 Astra Law in the model picker. Access is limited to eligible lawyers and people working under their supervision. The company has not announced a timeline for broader availability.
Plugins that connect firm tools to ChatGPT
Partner-built plugins bring a firm's existing knowledge and software into the ChatGPT environment. With the iManage plugin, a lawyer can draft a negotiation brief in ChatGPT and save it directly to the matter file. The DeepJudge plugin allows comparison against prior deals. These integrations aim to reduce the friction of switching between research, drafting, and document management systems.
Community plugins add skills built by lawyers and legal engineers at LegalQuants, LECG, and Skills.law. Firms can adapt these skills for their own workflows. OpenAI directs users to the Plugins section in ChatGPT and Codex for installation instructions, noting that each plugin's listing specifies its availability and access requirements.
Why this matters for product development in legal
For product teams inside law firms and legal tech companies, Astra for Law signals a shift toward domain-tuned models that come with embedded, continuously updated search indexes. The combination reduces the need to build and maintain separate retrieval pipelines for U.S. legal sources. The plugin architecture also provides a template for connecting proprietary firm data-matter files, deal history, document repositories-directly to a large language model without leaving the chat interface. Teams evaluating build-versus-buy decisions for AI for Legal research tools now have a baseline to measure against, particularly on data freshness and source attribution requirements. For paralegals and support staff who will operate these tools, structured upskilling through an AI Learning Path for Paralegals can close the gap between model capability and daily practice.
Your membership also unlocks:
