Article

Astra for Law access and pricing for legal teams

Law firms and in-house teams can compare access routes, approval requirements and API or Enterprise usage costs before adopting Astra for Law.

Editorial illustration for Astra for Law access and pricing for legal teams: a document represents the research briefing. Not documentary evidence.

On 17 November 2026, OpenAI opened Astra for Law access requests to every ChatGPT Enterprise customer, giving law firms and in-house legal teams a wider route to apply—not automatic approval. Software builders can enquire separately about API access. Current short-context Standard API rates are $8 per million input tokens and $40 per million output tokens; those usage charges are not a complete Enterprise subscription quote.

The updated offering pairs GPT-6.2 Astra with a legal search index and instructions for legal research and writing. The purchasing decision has three parts: whether your organization can obtain approval, which staff may use it, and how their work will be billed.

Who is eligible to use Astra for Law?

OpenAI’s access requirements distinguish permission to apply from permission to use the model. Only lawyers and people working under their supervision may use it, in eligible geographies, after Trusted Access approval.

The expansion does not establish an entitlement for a personal ChatGPT subscription. Nor does the public API enquiry form settle whether every API applicant must hold an Enterprise contract. Ask OpenAI to confirm your organization’s route before purchasing a plan specifically for Astra for Law.

Country eligibility and approval timing remain questions for the account team. OpenAI’s announcement describes additions to its legal search index from the UK, Australia and Canada; coverage of a country’s legal sources is not confirmation that customers in that country can obtain access.

How to request workspace or API access

For an Enterprise workspace, OpenAI’s documented application process is:

  1. Have an owner or admin open Admin Console → Trusted Access → Astra for Law.

  2. Verify the business, complete the form, sign the Legal Technology and Trust Addendum, and submit the application.

  3. After approval, activate Astra for Law under Admin Console → Models → Legal configuration.

  4. Assign authorized users through role-based access control—the settings that determine which staff can use it.

For a custom application, contact your OpenAI account team or Legal Sales. The public form includes a developer-interest option for the Astra for Law API. Submitting it is an enquiry, not confirmation that the model has been enabled for your account.

What are the API prices?

The current API price list identifies the new configuration as gpt-6.2-astra-law. Standard rates below are in US dollars per million tokens, the units used to measure model input and output.

Input size per request

Uncached input

Cached input

Output

Up to 272,000 tokens

$8

$0.40

$40

More than 272,000 tokens

$16

$0.80

$60

The context tier depends on each request, not the month’s combined usage. Cache writes are priced separately, and regional-processing endpoints carry a 10% uplift where applicable.

OpenAI commits to promotional pricing at least through 31 December 2026. That wording does not announce a price rise on 1 January or establish what later rates will be.

Which Enterprise billing model applies?

Enterprise usage prices are also public, but the applicable agreement matters. Do not assume that an API price or a row in a rate card grants access.

  • USD token-based agreements: the Enterprise rate card lists GPT-6.2 Astra Law in Work and Codex at $8 input, $0.40 cached input and $40 output per million tokens in Standard mode. These rates apply to eligible agreements; a workspace moving from credits must first change billing model.

  • Credit-based agreements: the credit rate card lists 200 input credits, 10 cached-input credits and 1,000 output credits per million tokens. These are consumption units, not a universal dollar invoice.

The legal configuration’s rate-card footnote specifies Standard mode only at launch, without Fast or Ultrafast. Do not transfer the general Astra model’s speed options to Astra for Law.

The USD Enterprise card also applies long-context multipliers above 272,000 input tokens, except in Codex when signed in with ChatGPT. That exception does not apply to API-key billing.

Confirm the subscription commitment, applicable discounts, included usage and additional feature charges with the contract owner. For the broader budgeting distinction, see our guide to Enterprise token billing and total agent costs.

An illustrative API budget for 100 requests

Suppose a pilot makes 100 requests, each with 100,000 uncached input tokens and 10,000 total billable output tokens. These are assumed volumes, not measured legal workloads. At the short-context Standard rates:

  • Input: 10 million tokens × $8 per million = $80.

  • Output: 1 million tokens × $40 per million = $40.

  • Total model usage: $120.

This excludes tools, cache writes, regional uplifts, tax, subscriptions, integration and lawyer review. It is not a quote for 100 completed legal matters.

Include reasoning in the output estimate: OpenAI bills reasoning tokens as output even though they are not visible in the API response. In this example, output accounts for one-eleventh of the tokens but one-third of the model bill. A short final answer alone is therefore a poor guide to cost.

What to confirm before committing

Obtain confirmation of geographic eligibility, authorized users, approval timing and the applicable commercial terms. For an integration, also establish which legal-search tools are provided and how their use is charged; the public pricing material does not isolate a legal-search-specific call tariff.

OpenAI describes its legal offering as providing zero data retention for API usage and excluding Enterprise usage from human review by default. Confirm how those commitments apply to your agreement and connected tools; they are not an audit of a particular deployment.

For a proposed pilot, choose an authorized, representative task and a supervising lawyer to check answers and cited sources. Record corrections, review time and the full cost before deciding to expand. Agree who can approve or stop further pilot spending before the next commitment.

For an existing Enterprise team, the next step is an admin-led access request. For a software builder, it is an API enquiry with a defined use case. In both cases, wait for access and contract confirmation before treating the published usage rates as a deployment budget.