Breaking

US Government to Approve GPT-5.6 Access Customer by Customer as OpenAI Limits Preview

June 25, 2026 at 17:15 EDT

Policy & Regulation
Foundation Models
Security

The Trump administration has asked OpenAI to stagger the release of its next model, GPT-5.6, introducing an unusual arrangement in which the government approves access on a customer-by-customer basis. According to reports dated June 25, 2026 (local time), OpenAI CEO Sam Altman told staff that the model would be released as a limited preview to a small group of partners, and that during this preview period the government would be "approving access customer by customer." Reuters reported that a broader public release is expected several weeks later.

Frontier AI · Government Review

GPT-5.6 May Ship Only as a Preview — Approved "Customer by Customer" by the U.S. Government

According to reports, OpenAI's next frontier model would reach only a small group of partners, with federal sign-off required per customer. Sam Altman is said to have called the arrangement "not a desirable model over the long term."

Per-customer

Government approval granted one client at a time

ChatGPT Enterprise to federal agencies via the OneGov deal

Official GPT-5.6 announcements — only rumors circulate

CyberGym score — GPT-5.5 versions

The dedicated Cyber build edges out the standard model (% on CyberGym).

81.8%

GPT-5.5
Standard

85.6%

GPT-5.5
Cyber

GPT-5.5 vs. GPT-5.6 (per reports)

GPT-5.5 — current frontier

Officially announced (April 23)
General release + government early access
CyberGym 85.6% (Cyber version)
Release timing confirmed

GPT-5.6 — rumored

No official announcement
Limited preview + per-customer approval
Benchmarks undisclosed
Staggered / delayed rollout expected

Toward a de facto licensing path

Pre-release review order

→

Government national-security testing

→

Per-customer access sign-off

"The government would approve access customer by customer — not a desirable model over the long term."

Reported context

Commerce Secretary said to have warned against launching without other agencies' sign-off
Some Anthropic models reportedly pulled temporarily at the government's request
Deepening ties: FedRAMP authorization, Trusted Access / Daybreak programs
Much remains based on reports and internal leaks — official confirmation is limited

Continue reading

The rest of this article is for AI News Blitz readers. Choose an option below to keep reading.