OpenAI is holding back the wider release of GPT-5.6 after a U.S. government request, putting its newest AI models into the hands of only a small group of approved partners for now.
The company said Friday that the preview is limited to partners whose participation has been shared with the government. The GPT-5.6 lineup includes Sol, OpenAI’s flagship model, Terra, a balanced model for everyday use, and Luna, a faster and cheaper option.
The restriction covers all three models, even though Sol is the most powerful system in the set. OpenAI said Sol has stronger agent-style abilities in coding, biology and cybersecurity, plus new reasoning modes designed for more complex tasks.
The decision follows fresh pressure from the Trump administration on frontier AI releases. After Anthropic made its Fable 5 model public, the administration ordered the company to remove access for foreign nationals, prompting Anthropic to take the model down entirely.
Dean Ball, a former White House AI adviser and incoming OpenAI employee, has argued that the administration’s review process creates what he called a de facto involuntary licensing regime for frontier AI. He said the lack of clear safety standards could lead to repeated launch delays.
OpenAI complied with the request, but pushed back on making the process permanent. “We don’t believe this kind of government access process should become the long-term default,” the company said.
The company said the limited preview is a short-term step while it works with the administration on a repeatable process for future model releases. OpenAI plans broader access through ChatGPT, Codex and its API in the coming weeks.
For developers and enterprise users, the rollout also comes with published pricing: Sol costs $5 per million input tokens and $30 per million output tokens, with Terra at half that price and Luna at $1 and $6 respectively.
