gpt-6-astra-law Is on the Pricing Page at Plain Astra's Rate, Labeled Promotional Through December 31. FedRAMP Endpoints Get a 10% Uplift From November 1.
OpenAI's API pricing page lists gpt-6-astra-law under legal model configurations at $10 input, $50 output, and $1 cached per million tokens, with a note that the promotional price holds at least through December 31, 2026. That matches plain gpt-6-astra, which implies a premium afterward that OpenAI has not published. The same page says that starting November 1, 2026, FedRAMP endpoints are charged a 10% uplift, alongside the existing 10% uplift for regional data residency on models released since March 5. Whether the law configuration still requires Trusted Access is unverified. We read the pricing page and the Astra for Law launch materials.
By T.W. Ghost
Two lines on OpenAI's API pricing page matter this week, and neither came with an announcement. The first: gpt-6-astra-law, the legal configuration of GPT-6 Astra that launched September 17 for selected US law firms under Trusted Access, is now listed under legal model configurations at $10 per million input tokens, $50 output, and $1 cached, with the note that "GPT-6 Astra for Law promotional pricing is available at least through December 31, 2026." Those are exactly the standard short-context rates for plain gpt-6-astra. A promotional price equal to the base price implies a higher price after the promotion, and OpenAI has not printed one.
The second line, on the same page: "Starting Nov 1, 2026, FedRAMP endpoints are also charged a 10% uplift." That sits next to the existing 10% uplift for regional data residency on models released on or after March 5, 2026. Two uplift families, both 10%, and the page does not spell out how they stack when a customer needs both. We read the pricing page, the Astra for Law launch post, and the help article. Here is what held up.
The Two Lines
| What the pricing page says | |
|---|---|
gpt-6-astra-law | $10 in / $50 out / $1 cached per 1M tokens; promotional "at least through December 31, 2026" |
gpt-6-astra (standard, short context) | $10 in / $50 out / $1 cached per 1M tokens |
| Post-promotion Astra for Law price | Not published |
| FedRAMP endpoints | 10% uplift starting November 1, 2026 |
| Regional data residency | 10% uplift, models released on or after March 5, 2026 |
| Stacking of the two uplifts | Not spelled out beyond both being 10% |
| Access to the law configuration | Launch said Trusted Access and selected US firms; current gating unverified |
A configuration, not a new model. Astra for Law is Astra with a legal setup. The pricing row makes that concrete: same token prices, different name, with a promotional label the base model does not carry.
Promotional at base price means the real price is higher. That is arithmetic about a word, not about a number. OpenAI's page gives the word and not the number.
FedRAMP is a second 10%, with a date. From November 1, a FedRAMP endpoint costs more than the same call on a standard endpoint. For a customer who also needs residency, the page gives two 10% lines and no combined example.
Verified vs Unconfirmed: The Scorecard
| Claim | Verdict |
|---|---|
gpt-6-astra-law at $10 / $50 / $1 per 1M | Verified (OpenAI pricing page) |
| Promotional at least through Dec 31, 2026 | Verified (pricing page) |
Plain gpt-6-astra standard short-context at the same rates | Verified (pricing page) |
| FedRAMP endpoints 10% uplift from Nov 1, 2026 | Verified (pricing page) |
| Regional residency 10% uplift for models from Mar 5, 2026 | Verified (pricing page) |
| Post-promotion Astra for Law price | Unverified (not published) |
| Trusted Access still required for the law configuration | Unverified (launch said selected firms) |
| How the two uplifts combine | Unverified (not spelled out) |
What To Do
Budget FedRAMP and residency as separate 10% lines before Monday, and ask OpenAI for a combined quote if you need both; the page will not give you one.
Treat Astra for Law as a promotional rate with an unknown step-up after December 31, and confirm your firm's access eligibility separately from the public price row.
The frontier is elsewhere: on the government side, Anthropic cut 40% off Claude tokens for agencies the same week OpenAI added 10% to FedRAMP endpoints; the two moves are not the same thing, and both are dated. Our model comparison has the current standings side by side.
Sources
*Not sure whether a promotional legal configuration or the base model is the right line item? Take the free 2-minute quiz and get matched. Then read about the government discount that runs the other direction: Anthropic's OneGov deal.*