Observed Signal · Nov 17, 2026 · Technical Release · Source: OpenAI Blog · Impact: 4/5 · Sentiment: Positive

OpenAI Astra for Law tops legal benchmarks

Executive Signal Summary

OpenAI released an updated Astra for Law, a model configuration of GPT-6.2 Astra, achieving state-of-the-art results on Harvey LAB-AA and top performance on LAB: USCL and Vals AI's Legal Research Bench. The tool is now available to all ChatGPT Enterprise customers via the Trusted Access Program, including API access. Astra for Law ranked first on LAB-AA with 20.8% full credit, ahead of GPT-6.2 Astra and non-OpenAI models, while using fewer output tokens. It also demonstrated 95% fewer hallucinations per assignment compared to Claude Opus 5.5. The legal search index expanded by 34% since September, now covering over 310 million URLs, including international jurisdictions. Early adopters like Cooley, Ropes & Gray, and Sullivan & Cromwell are using it for legal research, brief writing, and workflow automation.

Polaris7 AgentPolaris7 Strategic Assessment
High Confidence

OpenAI's release of a state-of-the-art legal AI model marks a significant advancement in AI applications for the legal industry, potentially transforming legal research and practice, and signaling a competitive edge in the AI model landscape.

MARKET SIGNALS

Track Harvey Signals & Strategic Shifts

Polaris7 connects regulatory filings, primary sources, leadership changes, and M&A activity into structured intelligence. Set up your Polaris7 workspace to monitor relevant companies in your executive briefing.

  • Verified primary sources
  • Automated briefings
  • Instant watchlist setup

Key Takeaways & Evidence Grounding

  • OpenAI released updated Astra for Law, built on GPT-6.2 Astra.
  • Astra for Law ranked first on Harvey LAB-AA with 20.8% full credit.
  • Astra for Law showed 95% fewer hallucinations per assignment than Claude Opus 5.5.
  • Legal search index expanded by 34% to over 310 million URLs.
  • Astra for Law is now available to all ChatGPT Enterprise customers via Trusted Access Program.

Connected Companies & Entities

“Harvey's Legal Agent Benchmark (LAB) tests whether a model can take a legal assignment from start to finish.”

Ontology Mapping & Concepts

Primary Source Grounding & Direct Attribution
Direct Origin Attribution
Primary Reporting: OpenAI Blog•Published: Nov 17, 2026
Original Coverage Title: “Astra for Law leads in legal intelligence”

Related Market Signals & Shifts

Recent verified developments and strategic activity across this market segment.

AISep 17, 2026

OpenAI Launches Astra for Law for Legal AI

OpenAI has introduced Astra for Law, a specialized foundation for law firms and legal technology companies to build AI products and workflows. It combines GPT-6 Astra with a legal search index, custom instructions, and tools tailored for legal work. The offering includes a Trusted Access Program with zero data retention, and partnerships with law firms like Latham & Watkins for governance. Eighteen new ecosystem plugins connect ChatGPT to tools like Relativity and Clio. Astra for Law is initially available to selected firms through Trusted Access in ChatGPT and Codex, with API access coming soon. OpenAI also highlighted performance improvements in legal research and end-to-end workflows, and announced community plugins and ChatGPT for Word.

Read assessment
AI in Legal TechSep 23, 2026

Harvey leverages GPT-6 Astra to improve legal document drafting

OpenAI highlights Harvey, an AI legal platform, using GPT-6 Astra to enhance legal document drafting. Harvey processes extensive legal context, such as court records and case law, to produce more structured and complete outputs. The integration allows lawyers to encode preferences via a memory panel, influencing the drafting process. This collaboration demonstrates the application of advanced AI in legal workflows, improving efficiency and output quality.

Read assessment
PlatformSep 3, 2026

OpenAI begins rolling out GPT-6 Astra model

OpenAI's GPT-6 Astra, launched on September 4, 2026, advances toward AGI, tying Claude Fable 5.1 at 53 points on the Artificial Analysis Intelligence Index v4.3 and leading new benchmarks like Terminal-Bench 4.0 and AutomationBench-AA. It matches Claude on the Coding Agent Index at less than half the cost, but independent ARC-AGI-3 scores vary widely, and safety concerns persist. The company also showcases Astra's engineering capabilities: it designed a complete device without human edits, expanded partnerships with Chip Ganassi Racing and Mazda, and introduced BenchCAD 2.0, where a newer internal model outperforms Astra. OpenAI serves over one billion weekly users and plans to deploy its Jalapeño chip by year-end, while aiming to reclaim enterprise coding share from Anthropic.

Read assessment

WATCHLIST & BRIEFINGS

Receive future signals for Harvey & competitors in your briefing

Stay ahead of material M&A deals, leadership changes, account wins, and product rollouts. Polaris7 connects primary filings, company newsrooms, and trade reporting directly into your executive briefing.

  • Regular primary-source checks
  • Structured Watchlist briefings
  • Preserves current context