OpenAI GPT-5.6 cleared for government

OpenAI’s GPT-5.6 models are now available for U.S. government work through the company’s FedRAMP-authorized ChatGPT Enterprise channel, according to Nextgov, ending a weeks-long period in which the models were limited after White House intervention. Microsoft, meanwhile, has told developers working on internal AI coding projects to use GPT-5.6 Sol as the default model in GitHub Copilot, CNBC reported on August 5. A separate Washington Post report said White House officials briefed tech companies on Tuesday on a framework that would exempt “open” AI systems from security review while allowing scrutiny of cutting-edge closed systems. ### How did GPT-5.6 get from restricted rollout to government availability? OpenAI began limiting GPT-5.6 access on June 26 after a U.S. government request that the initial release be confined to a “small group of trusted partners,” CNBC and Nextgov reported at the time. OpenAI later said it was working with senior Trump administration officials ahead of a broader deployment, and Nextgov reported on July 8 that the model’s availability followed government evaluation tied to a June AI executive order. (nextgov.com) August 4 brought the next step. Nextgov reported that GPT-5.6 had gone live for government use through FedRAMP-authorized ChatGPT Enterprise, weeks after one of the advanced models had been linked to an autonomous digital attack scenario that intensified official scrutiny. The same report said the channel gives agencies a procurement path that fits existing federal cloud compliance requirements. (cnbc.com) ### Why does the FedRAMP channel matter more than a normal product launch? FedRAMP is the federal program used to authorize cloud services for government use, and OpenAI had already been building a government-specific path through products such as ChatGPT Gov and agency access agreements. Nextgov reported in 2025 that OpenAI offered federal agencies ChatGPT access through a OneGov pact, and in January 2025 the outlet reported the launch of ChatGPT Gov for secure cloud environments. (nextgov.com) June 16 added another signal. Nextgov reported that OpenAI was working with the Pentagon’s Chief Digital and Artificial Intelligence Office to bring ChatGPT access to GenAI.mil in early July. That means the August 4 availability of GPT-5.6 is not a stand-alone release; it extends an existing government distribution channel into newer frontier models. (nextgov.com) ### Why is Microsoft standardizing on GPT-5.6 Sol internally? Microsoft told developers this week to default to GPT-5.6 Sol for AI coding work in GitHub Copilot, according to a memo viewed by CNBC. Jay Parikh, executive vice president of Microsoft’s CoreAI engineering group, wrote that “internally, shifting more workloads to OpenAI models helps us get greater value from our token investment,” CNBC reported. (nextgov.com) June 2 showed why that instruction stands out. CNBC reported then that Microsoft had been promoting its own MAI-Code-1-Flash model as part of a push to reduce dependence on OpenAI and lower costs for developers. The August 5 internal directive therefore shows Microsoft still using OpenAI’s top model as the default option for at least some staff workflows, even as it builds competing models. (cnbc.com) ### What exactly is the White House doing on model reviews? The Washington Post reported on August 4 that White House officials briefed tech companies on a framework that would exempt “open” AI systems from security review while allowing officials to vet cutting-edge systems for cybersecurity risks. Wired separately reported that OpenAI delayed the GPT-5.6 rollout in response to a White House request, and CNBC reported in June that the staggered release followed a Trump administration review request. (cnbc.com) July 9 added a public marker. CNBC reported that Sam Altman said OpenAI had worked with Commerce Secretary Howard Lutnick, Treasury Secretary Scott Bessent and U.S. officials ahead of deployment of GPT-5.6 Sol, Terra and Luna. That places model access inside a policy process, not just a product timetable. ### What does this change for companies building on frontier models? (washingtonpost.com) August 5’s developments leave enterprise buyers with two concrete signals. OpenAI now has a government-cleared path for GPT-5.6 through a FedRAMP-authorized enterprise wrapper, and Microsoft is treating GPT-5.6 Sol as a sanctioned default for internal coding work. At the same time, the White House framework described by The Washington Post would preserve extra review for some advanced closed systems while exempting “open” ones. (cnbc.com) The next milestones are already visible. The White House framework was presented to companies on Tuesday, August 4, according to The Washington Post, while federal agencies can now reach GPT-5.6 through ChatGPT Enterprise, Nextgov reported, and Microsoft staff are already being directed to use Sol inside GitHub Copilot, CNBC said. (nextgov.com)