OpenAI · 2026-09-17 · major
Astra for Law — OpenAI ties GPT-6 Astra to a 230M-URL legal index
Astra for Law is OpenAI's legal setup for GPT-6 Astra: a search index of more than 230 million URLs of US law plus 26 partner plugins. OpenAI reports 54% accuracy on 200 legal research questions, against 38.7% for GPT-6 with web search.

OpenAI's GPT-6 Astra, wired to a searchable index of US law and 26 legal vendor plugins.
Quick facts
| Maker | OpenAI |
|---|---|
| Base model | GPT-6 Astra |
| Legal search index | 230M+ URLs |
| What it covers | US caselaw, statutes, regulations, court rules, administrative decisions |
| Plugins | 26 vendor, 9 community, 47 skills |
| Availability | API, plus a Trusted Access Program for selected Am Law 200 firms |
| Data handling | Zero data retention, excluded from human review |
Benchmarks
What is it?
Astra for Law adds a dedicated legal search index to GPT-6 Astra — more than 230 million URLs of US caselaw, statutes, regulations, court rules and administrative decisions, drawn from the nonprofit Free Law Project's CourtListener library. OpenAI announced it on 17 September 2026 as an offering for law firms and legal technology companies rather than a new model. Custom instructions for legal analysis and writing ship with it.
How does it work?
Retrieval is the change. Instead of searching the open web, the model queries the curated legal corpus, and OpenAI pairs that with a plugin layer: 26 plugins built by vendors including Thomson Reuters, Intapp, Harvey, Legora, DeepJudge and iManage, 9 community plugins and 47 customizable skills. Firms reach it through the API or a Trusted Access Program, with zero data retention and no human review of the traffic.
Why does it matter?
Citation errors are the reason legal teams have kept general chatbots out of billable research, and a curated index is a direct answer to that. OpenAI's own numbers put Astra for Law at 54% on a 200-question legal research test against 38.7% for GPT-6 with web search — still far from reliable, but enough of a jump that vendors like Harvey and Legora are building products on it rather than on plain web search.
Who is it for?
law firms and legal tech builders
Frequently asked questions
- How much does Astra for Law cost?
- OpenAI did not publish prices for Astra for Law in the launch post. Access runs through two paths instead of a listed plan: the API, which customers such as Harvey and Legora can build on, and a Trusted Access Program aimed at selected Am Law 200 firms. Firms and vendors negotiate terms directly rather than picking a published tier.
- Who can use Astra for Law right now?
- Astra for Law is open to API customers, with Harvey and Legora named as builders on it, and to law firms admitted to OpenAI's Trusted Access Program, which targets selected Am Law 200 firms. It is not a consumer ChatGPT feature. OpenAI says it will bring the same legal capabilities to later frontier models as they ship.
- How is Astra for Law different from asking GPT-6 Astra with web search?
- Astra for Law swaps open web search for a dedicated legal index and adds instructions written for legal analysis and drafting. On OpenAI's test of 200 legal research questions, Astra for Law answered 54% correctly against 38.7% for GPT-6 with web search, and it retrieved 24% more relevant cases and up to 54% more relevant passages.
- Where does the legal data behind Astra for Law come from?
- The Astra for Law index is built on CourtListener, the legal research library run by the nonprofit Free Law Project, and spans more than 230 million URLs of US caselaw, statutes, regulations, court rules and administrative decisions. That is a curated legal corpus rather than a general web crawl, which is what the accuracy gap is credited to.
- Does OpenAI keep what law firms send to Astra for Law?
- OpenAI offers zero data retention on Astra for Law and excludes the traffic from human review. LawSites reports those terms arrive as a roughly 30-page agreement with exceptions that are not spelled out publicly, so firms handling privileged material still have to read the contract rather than rely on the headline promise.