OpenAI Inc.'s CEO Sam Altman stated on Thursday that the company's newest AI model, GPT-5.6, was 54% more token-efficient for performing agentic coding. Speaking to CNBC, Altman emphasized that GPT-5.6 was "as good or better" than similar models from other companies, but acknowledged a present rise in the cost of compute and memory, which he called "definitely a headwind" for the sector. Altman also commented on an earlier limited release of the model for several federal agencies, describing the cybersecurity work with the government as a "collaborative back and forth." He added, "If you want broad access, which we do, and you have powerful models, you really want to be able to be confident in your safety claims, because otherwise the world is going to get uncomfortable very fast."