OpenAI’s GPT-5.6 cleared for public launch Thursday after federal review

OpenAI will publicly release its GPT-5.6 model lineup — including its flagship Sol, alongside Terra and Luna — on Thursday, July 9, after the U.S. Department of Commerce approved a broad rollout following weeks of restricted access.aiweekly+2

"GPT-5.6 Sol, along with Terra and Luna, will launch publicly this Thursday. We're expanding preview access globally now," the company said in a statement.khaleejtimes+1

The approval, first reported by Axios, marks the conclusion of a federal review process conducted by the Commerce Department's Center for AI Standards and Innovation (CAISI), the first real-world test of President Trump's June executive order establishing a voluntary framework for pre-release evaluation of frontier AI models.letsdatascience+1

From Restricted Preview to Public Release

OpenAI initially previewed GPT-5.6 on June 26 but limited access to roughly 20 government-approved partners after the Trump administration raised cybersecurity concerns about the model's capabilities. The White House's Office of the National Cyber Director and Office of Science and Technology Policy had asked OpenAI to restrict the rollout while a federal testing framework was developed.axios+3

Commerce Secretary Howard Lutnick wanted assurance that relevant government agencies had tested and reviewed the model before a wider release, according to Axios. To expedite the process, OpenAI stationed technical experts in Washington to field inquiries throughout the evaluation period.axios+1

The staggered release mirrored a parallel process undergone by Anthropic, whose own frontier models Mythos and Fable faced similar restrictions before gaining broader clearance.aiweekly

What GPT-5.6 Brings

OpenAI describes Sol as a "step function better than GPT-5.5," with advances in coding, scientific reasoning, long-horizon planning, and agentic workflows. On cybersecurity benchmarks, Sol matched the performance of Anthropic's Mythos Preview while using roughly one-third of the output tokens. In biology, it scored about nine percentage points above GPT-5.5 on SecureBio evaluations.openai+1

The model family is priced per million tokens: Sol at $5 input and $30 output, Terra at $2.50 and $15, and Luna at $1 and $6. Sol also introduces "max" reasoning effort and an "ultra" mode that deploys coordinated subagents for complex tasks.openai+1

A New Regulatory Precedent

The episode has established a working template for how Washington and AI developers interact before major launches. The executive order's framework is nominally voluntary, but the GPT-5.6 process suggests it is quickly becoming a de facto requirement for frontier releases. Whether smaller labs and non-U.S. developers will face comparable scrutiny remains an open question.axios+1