Users anticipating the arrival of OpenAI’s new GPT-5.6 AI models will have to wait a little longer. According to a person familiar with the company's internal perspective, OpenAI is not exactly thrilled with the current situation but views the delay and the associated government approval process as merely temporary. In an official blog post, the company expressed optimism that it would be able to make GPT-5.6 available to the public in the coming weeks. Reports regarding OpenAI's plan to postpone its next generation of AI models at the request of the Trump administration have previously circulated.
Government Access and Long-Term Concerns
In its blog post, OpenAI stated that it does not believe this government access process should become a long-term standard. The company argues that such procedures prevent the best tools from reaching the developers, enterprises, cybersecurity defenders, and global partners who rely on them. OpenAI explained that it is taking this short-term step as it believes it is the most effective path toward broader availability in the upcoming weeks, while concurrently working with the Administration to develop an Executive Order framework for cybersecurity and a repeatable process for future model releases.
The Executive Order and Cybersecurity
Earlier this month, President Donald Trump signed an executive order aimed at addressing cybersecurity risks posed by powerful new AI models. The order proposed a voluntary process for AI labs to share their models with the government 30 days before a wider release. While the mandate included a carve-out stating the US government would not turn this voluntary process into a de facto licensing regime for model releases, OpenAI executives noted during a Friday briefing that no such voluntary framework currently exists. Consequently, frontier AI labs find themselves in an awkward interim phase, where collaborating with the US government on an AI launch feels far from truly voluntary.
The Standoff with Anthropic
The White House request for OpenAI to stagger its AI releases follows just two weeks after an export control directive was sent to Anthropic, which forced the company to take its most advanced AI models offline for all customers. The dispute between Anthropic and the White House remains unresolved, and some of the company’s own employees are still prohibited from accessing its most advanced systems.
Impact on US AI Innovation
The Trump administration’s move to limit the availability of advanced models at OpenAI and Anthropic has created an environment of uncertainty for other US-based AI labs. Over the past two years, the administration has focused on clearing regulations and red tape that could hinder American AI innovation and weaken the country's competitive standing against China. However, in recent months, the White House has grown increasingly concerned about the cybersecurity implications of new models and is scrambling to address those risks. A White House official confirmed that the administration continues to collaborate with these frontier AI labs.
The Three Flavors of GPT-5.6
OpenAI plans to broaden the customer base for GPT-5.6 starting next week, including various international partners. Company executives explained that they cannot disclose the specifics of how the White House approves these customers, noting that the process primarily involves submitting a list to the US government and awaiting feedback. OpenAI revealed that GPT-5.6 will be offered in three distinct versions: Sol, the most capable iteration; Terra, a mid-tier version; and Luna, a faster and more affordable option. OpenAI claims that GPT-5.6 Sol is its most advanced model yet in benchmarks testing cybersecurity, biology, and agentic capabilities. To accompany these features, the company has implemented a layered safeguard stack intended to prevent malicious actors from utilizing the model for cyberattacks or other harmful behaviors.
