OpenAI is postponing the full public release of GPT-5.6 at the request of the US government. Initially, access will be limited to a small group of vetted partners.
The company announced this decision on June 26, 2026 in a blog post, explaining that the limited launch is a temporary measure while it works with Washington to develop a broader framework for future launches.
Details about the partners involved were shared with US authorities before publication. However, the names of these partners have not been made public.
This decision affects OpenAI’s new lineup of models, including the GPT-5.6 Sol, which is the most advanced, the GPT-5.6 Terra, a mid-level option, and the GPT-5.6 Luna, which is designed to be more affordable.
What the US government is doing and how OpenAI is responding
The delay comes after an executive order signed by President Donald Trump earlier this month. The order creates a voluntary framework for AI developers to offer “covered frontier models” to the US government for up to 30 days before turning them over to trusted partners.
The goal is to give U.S. officials early access to powerful artificial intelligence systems so they can identify potential national security risks, including cyberattack capabilities and military misuse, before the tools become widely available.
OpenAI shared its plans and the models’ capabilities with the US government ahead of the limited launch.
OpenAI explained: “We are taking this step in the short term because we believe it is the best way to increase availability in the coming weeks, while we work with the Administration to develop the Cyber Executive Order framework and a process for future model releases.”
The company also clarified that it does not want the current level of government oversight to become permanent. OpenAI expressed concern that the process could limit access to advanced AI tools for developers, businesses, cybersecurity professionals and international partners who could benefit from them.
CEO Sam Altman addressed the situation at X, saying that extensive security testing “isn’t a bad idea. I just don’t like the idea of the government choosing customers.”
Context
The GPT-5.6 delay fits into a broader pattern of US government involvement in advanced AI models. Recently, Anthropic discontinued the availability of Claude Fable 5 and Mythos 5 after a US government order restricted access by foreign nationals over fears the models could be exploited to bypass security measures.
Anthropic and numerous cybersecurity experts have criticized these restrictions, arguing that the mentioned vulnerabilities are common among many publicly available models.
The Pentagon previously called Anthropic a “supply chain risk” because the company refused to allow its models to be used for autonomous weapons or mass domestic surveillance. A federal judge ruled that the Pentagon directive is unenforceable while Anthropic’s lawsuit is ongoing.
The growing tension between AI companies and the US government highlights ongoing debates over how to regulate powerful AI systems before they are widely deployed.
What the GPT-5.6 delay means for users and what comes next
GPT-5.6 will not be available to general users, including ChatGPT Plus, Pro, and Enterprise subscribers, until the limited trial period with vetted partners ends. OpenAI has suggested that wider availability could come in the coming weeks, but has not given a specific date.
Current OpenAI models, such as GPT-5 and earlier versions, remain accessible through existing channels. In the meantime, users who need cutting-edge model capabilities may need to rely on current models or alternative services until GPT-5.6 is publicly released.
The delay applies specifically to GPT-5.6. Future releases of the OpenAI model will depend on the framework established through ongoing discussions between OpenAI and the US government.
OpenAI and the US government are working together to develop a repeatable process for future model releases. This framework is expected to describe:
- What AI capabilities do government access requirements enable?
- The length of pre-launch government access periods
- The information AI developers must provide
- What government agencies will have access?
While the current framework is voluntary and negotiable, it also introduces uncertainty about how restrictive future rules could become.
OpenAI’s confidential S-1 IPO filing and Anthropic’s ongoing IPO efforts could also influence its engagement with regulators in the coming months.
The limited release of GPT-5.6 is now underway. OpenAI has not announced a specific date for the public launch, but expects wider availability in the coming weeks.
