OpenAI CEO Sam Altman said on Thursday that the company's latest artificial intelligence model, GPT-5.6 Sol, improved labeling efficiency by 54% in agent programming tasks and "performed as well or better than competing products on the market."

"Every business today is thinking about spending and the return they get from AI, and that's what we really want to do," Altman said.

OpenAI on Thursday launched its latest series of models - GPT-5.6 Sol, Terra and Luna, which were announced last month. At the request of the U.S. government, the company will initially only conduct a limited release to a "small group of trusted partners."

Altman said the company worked with Commerce Secretary Howard Lutnick, Treasury Secretary Scott Bessant and National Cyber ​​Director Sean Cairncross during the approval process. He described the collaboration between OpenAI and the government as a process of "back-and-forth communication and joint advancement." The government will conduct tests and raise issues, and the company will be responsible for solving them.

"If you want widespread use, which we certainly hope, and you have a robust model, you have to be confident in your security claims or the world will get upset very quickly," Altman said.

Previous reports stated that OpenAI is in preliminary and ongoing negotiations with the Trump administration to discuss the government’s possible stake in the company. According to the Financial Times, OpenAI has proposed to sell 5% of its shares to the government, but Altman said that "there are many inaccuracies in it."

Altman said he hopes the regulatory approach will be global and that people will be able to use AI without worrying about safety issues.

“Everyone has access,” he said, “and the United States does not benefit disproportionately.”


OpenAI was founded in 2015 as a non-profit research organization, and entered the mainstream after launching the ChatGPT chatbot in 2022. The company is currently valued at $852 billion by private investors and has sparked a race for AI dominance with rivals Anthropic, Google, Microsoft, Amazon, Meta and Elon Musk's SpaceX.

Meta released its latest AI model, Muse Spark 1.1, on Thursday, calling it "the most powerful agent and programming model to date." SpaceX acquired Musk's AI startup xAI earlier this year and launched a model called Grok 4.5 on Wednesday.

SpaceX completed its largest-ever IPO last month, and OpenAI and Anthropic are also preparing for potentially huge public market debuts. Both companies have confidentially filed prospectuses with regulators but have not announced formal plans or timetables.

When asked whether OpenAI would IPO this year, Altman remained tight-lipped.

"I don't know," he said.