Advancements in OpenAI Model Capabilities
During a recent appearance at Dreamforce 2026, OpenAI CEO Sam Altman provided specific benchmarks regarding the mathematical reasoning capabilities of the company’s evolving model lineage. According to statements captured from the session featuring Marc Benioff, Altman categorized the performance of various iterations of OpenAI’s technology through the lens of academic expertise.
Altman noted that GPT 5.5 performed at a level comparable to an average mathematics professor. He further stated that GPT 5.6 demonstrated capabilities equivalent to a professor in the top one or two percentile. While the Astra model was described as slightly superior to those benchmarks, Altman claimed that an undisclosed internal model has surpassed Astra, possessing the ability to perform tasks that “the best mathematicians in the world cannot.”
These claims, shared via social media platforms such as X and discussed on Reddit, have met with varying degrees of skepticism from observers. Some commenters suggested that while human expertise improves slowly, AI development follows a more consistent and rapid trajectory. Others cautioned that such high-level claims from a CEO may be subject to embellishment, noting that the truth likely exists in a nuanced middle ground between technological optimism and extreme skepticism. As these performance metrics have not been independently verified by third-party academic institutions, they remain unconfirmed.
IPO Outlook and Safety Concerns
In a separate development regarding the company’s corporate trajectory, Sam Altman addressed the timeline for a potential Initial Public Offering (IPO). Altman stated that an OpenAI IPO will not occur in 2026, citing ongoing concerns regarding AI safety.
The decision to delay a public listing comes amid a complex regulatory and social landscape regarding the deployment of advanced artificial intelligence. While some market observers have expressed anticipation for a public offering, the current focus for the company appears to be centered on addressing the safety implications of its increasingly capable models.
The intersection of rapid AI advancement and corporate governance continues to be a point of intense scrutiny for investors and regulators alike. As OpenAI navigates these safety considerations, the timeline for its transition to a public entity remains uncertain.