OpenAI will publicly launch all three GPT-5.6 variants, Sol, Terra, and Luna, on Thursday, July 9, after the Trump administration approved a wider rollout following additional government testing. The company announced the expansion on X Tuesday night, writing “We’re expanding preview access globally now,” according to Engadget.
The launch follows an unusual process. President Trump signed an AI cybersecurity executive order in early June asking companies to voluntarily present their most powerful models for government review 30 days before public release. OpenAI initially complied by limiting GPT-5.6 access to a small group of government-approved partners in late June. The Department of Commerce’s Center for AI Standards and Innovation then conducted additional testing, with OpenAI sending technical experts to Washington to address questions in real time, according to Axios reporting cited by Engadget.
The approval came well short of the 30-day window. OpenAI itself has signaled discomfort with the arrangement: “We don’t believe this kind of government access process should become the long-term default,” the company said when it first limited the release, but added that compliance was the fastest path to a public launch.
The GPT-5.6 Model Family
Sol is OpenAI’s new flagship, with the company claiming stronger agentic capabilities in coding, biology, and cybersecurity. Terra targets everyday use with performance comparable to GPT-5.5 at half the price. Luna is the lowest-cost option focused on speed and affordability.
Pricing breaks down as follows: Sol costs $5 per million input tokens and $30 per million output tokens. Terra runs $2.50/$15, and Luna comes in at $1/$6, according to Engadget.
Musk Responds Within Hours
Hours after OpenAI’s announcement, Elon Musk said SpaceXAI would launch Grok 4.5 “in the coming days,” describing it as an “Opus-class model, but faster, more token-efficient and lower cost,” according to Business Insider. The timing extends the public rivalry between Musk and OpenAI CEO Sam Altman, which intensified after Musk lost his lawsuit against OpenAI earlier this year. A jury ruled Musk had waited too long to file his claim that OpenAI betrayed its nonprofit agreement.
Google is also planning to release Gemini 3.5 Pro later this month, Business Insider previously reported.
The Regulatory Precedent
The broader pattern matters more than any single model launch. Anthropic faced a similar process: the Trump administration placed an export ban on its Mythos and Fable models over concerns about adversary misuse for cyberattacks. Anthropic suspended both models and has since received permission to redeploy Mythos 5, with Fable 5 returning last week, according to Engadget.
Both cases establish the same dynamic: frontier model releases now involve a government vetting step. For organizations building agent infrastructure on these models, this adds a new variable to deployment planning. Model availability is no longer purely a commercial decision by the lab. It depends on regulatory approval timelines that are negotiated, as Axios reports, “case-by-case, in real time.”