OpenAI CEO Sam Altman told CNBC on Thursday that the company’s latest artificial intelligence model, GPT-5.6 Sol, is 54% more token efficient in agent coding tasks, “on par or better” than competing models in the market.
“Every company right now is thinking about the spend and the value they’re getting in exchange for AI, and this is what we really want to do,” Altman said.
On Thursday, OpenAI will roll out its latest models, the GPT-5.6 Sol, Terra, and Luna series, which were announced last month. The company limited its initial launch to a “small group of trusted partners” at the request of the U.S. government.
Altman said the company worked with Commerce Secretary Howard Lutnick, Treasury Secretary Scott Bessent and U.S. National Cyber Secretary Sean Cairncross on the approval process. He described OpenAI’s work with the government as a “back-and-forth collaboration,” with the government conducting the tests and raising issues for the company to address.
“If you want broad access and you have a strong model, as we do, you really want to be able to have confidence in your safety claims. If you don’t, the world will quickly become uncomfortable,” Altman said.
As CNBC previously reported, OpenAI is in preliminary and ongoing talks with the Trump administration about a possible investment in the company. The Financial Times reports that the government is proposing to give the government a 5% stake, but Altman said “there’s a lot wrong with that.”
Altman said he hopes the regulatory approach will be global and allow people to use AI without worrying about safety.
“Everyone will have access,” he said. “The United States is not gaining any unfair advantage here.”
OpenAI was founded in 2015 as a nonprofit research company and broke into the mainstream in 2022 with the launch of its ChatGPT chatbot. Now valued at $852 billion by private investors, the company has embarked on a mad dash for AI supremacy among rivals such as: human, google, microsoft, Amazon, Meta and Elon Musk’s space x.
Meta announced Thursday its latest AI model, Muse Spark 1.1, which it says is its “most powerful model yet for agent and coding work.” SpaceX, which acquired Musk’s AI startup xAI earlier this year, on Wednesday launched a model called Grok 4.5.
SpaceX pulled off the largest IPO in history last month, and OpenAI and Anthropic are also gearing up for potentially big public market debuts. The companies have confidentially filed prospectuses with regulators, but have not disclosed formal plans or timelines.
Altman remained silent on whether OpenAI will IPO this year.
“I don’t know,” he said.
Spotlight: OpenAI CEO: China’s open source model is getting much better
