OpenAI unveils GPT-5.6 amid AI regulatory drama in the US

Less than 24 hours after news broke about OpenAI doing just that Reel the next model release At the request of the Trump administration, that model, GPT-5.6, is here. On Friday, the company revealed a limited preview of its new GPT 5.6 model range: Sol, the flagship; Terra, a mid-level model for “high-volume work”; and Luna, a “fast and affordable” daily model. OpenAI says it is particularly skilled in programming, cybersecurity, and biology, as well as maintaining focus during long-range AI missions.

Per million tokens, GPT-5.6 Sol is priced at $5 input/$30 output (roughly half the cost of Anthropic’s Claude Fable 5, which is $10 input/$50 output). Terra is half the cost of Sol, and Luna is less than half the cost of Terra. The company has also launched two additional modes for Sol: a “maximum” mode for deeper inference and a “ultra” mode to take advantage of subproxies – evocative of OpenClaw, and perhaps a sign of the creator of OpenClaw Peter SteinbergerWorked at OpenAI so far.

Unsurprisingly amid A Security scare in Washington, DCOpenAI has dedicated the majority of its advertising blog posts to safety and potential misuse. He seemed to be referring to Recent prison escape actions Of its competitor Anthropic, it wrote that “GPT-5.6 is trained to reject blocked cyber assistance, including when users attempt to hide their intentions or jailbreak a form.” It also said that Sol’s leading model is “better at helping people find and fix vulnerabilities than reliably performing full-scale attacks,” and that Sol does not cross a critical cyber threshold under OpenAI’s preparedness framework — though it’s worth noting that OpenAI Recently reviewed its preparedness framework in April and removed some areas of previous study.

The company said Sol has the company’s “strongest security suite to date” and that it has “enhanced protection for high-risk activities, sensitive cyber requests, and frequent misuse.” OpenAI said it has dedicated “approximately 700,000 GPU A100e hours” to the robotic red team and has also worked with third-party testers, the latter of which will continue testing over the next two weeks.

OpenAI also appears to be taking a hypersensitive approach during the preview period, which is being closely monitored by the Trump administration. “Safeguards may sometimes interfere with legitimate operation, especially in dual-use areas where defensive and offensive activity may initially appear similar. This is part of what the preview was designed to test,” the company wrote. The Trump administration will approve agents on a case-by-case basis during the preview period, the report said earlier this week.

OpenAI said the prototype should be generally available in the coming weeks because the company believes in “broad access,” and that the company has collaborated with the US government ahead of this launch, but that will hopefully not be the norm.

“We do not believe this type of government access process should become the default in the long term,” the company wrote. “It preserves the best tools from users, developers, enterprises, cyber defenders, and global partners who need them. We are taking this short-term step because we believe it is the strongest path to broader availability in the coming weeks, as we work with the administration to develop the Cyber ​​Executive Order framework and a repeatable process for future model releases.”