OpenAI CEO Addresses Rumors of ‘GPT-5 Orion’ Launching this Year

OpenAI CEO Sam Altman recently responded to speculation about a potential new Artificial Intelligence model, rumored to be called “Orion” and potentially representing ChatGPT-5, with a possible launch around December.

The story gained momentum after a subsequent report published exclusive claims about Orion that marked it as designed for the enterprise only and would only be available as an API to key partners. This controlled release strategy was described by one source as OpenAI’s plan to “grant access first to companies it works closely with in order for them to build their own products and features.” Another source also disclosed that Microsoft engineers are planning to roll out Orion as a part of the Azure beginning in November.

Altman, however, quickly responded to the report on X (formerly Twitter), calling it “fake news out of control.” His comment, which did not specify which parts were inaccurate, has led to ongoing debate. Altman did not outright deny Orion’s existence or dispute the release timing, making his response a nuanced refutation rather than a definitive dismissal.

The statement does not rule out the possibility that the rumors are inaccurate; however, it may also indicate that the company is uncertain about the timing of such decisions, as well as other variables. Speculation has often surrounded OpenAI’s upcoming releases, and projected timelines have sometimes shifted, surprising audiences.

In recent months, there has been speculation that OpenAI has been working on the development of GPT-5 and an AI-based Google search engine. Altman swiftly disavowed any claims of a new AI-centric search engine or GPT-5 being unveiled at the special event. However, he teased those who were working on an exciting project, potentially the impending GPT-4 Turbo.

It also asserts that it will disclose the safety testing procedures that purportedly caused stress exclusively for OpenAI’s safety testing group. During the same period, over half of the company’s key executives resigned, with some of them citing safety concerns in their resignation letters.

During the initial half of the year, there were models such as “OpenAI ROADMAP,” “o1 preview,” and “o1-mini,” which exhibited a distinctive “reasoning” architecture. Nevertheless, these releases failed to garner significant attention due to a variety of factors, including the application’s relative simplicity and high operating costs.

It was logical for OpenAI to have a high-performing agent like Orion in order to maintain competitiveness and capture the market’s attention, given that competitors like Anthropic were releasing similar innovations, such as their ‘Computer Use‘ agent and an enhanced Claude model.