As the most talked-about AI model, ChatGPT, approaches its second anniversary, rumors about OpenAI's new flagship model have once again sparked industry-wide discussions. However, this buzz was quickly dispelled by OpenAI's CEO, Sam Altman, himself.
The story originated from a report by technology media outlet The Verge. According to multiple insiders, OpenAI might unveil a new flagship model, codenamed "Orion," on November 30th, the second anniversary of ChatGPT. The report also mentioned that Microsoft Azure's engineering team has already begun preparations to host the new model. Unlike GPT-4 and GPT-4Turbo, this new model will be prioritized for partners, allowing them to develop related products and features first.
However, just as the tech community was rejoicing over the imminent arrival of "GPT-5," Altman quickly denied the rumors, pointing out the rampant spread of fake news. Nonetheless, he hinted at "many great things on the horizon," expressing dissatisfaction with the media publishing "random fantasies."
Interestingly, these "fantasies" might not be entirely unfounded. OpenAI's GPT-4Turbo, released in September (internal code name "Strawberry"), was considered to be generating training data for the next model. Given that over a year and a half has passed since the release of GPT-4, the development of a new generation model is quite logical.
Altman himself also posted a poem on social media depicting the winter night sky, which was interpreted as a hint of the new model's arrival. According to astronomical knowledge, Orion, one of the most prominent constellations in the northern hemisphere's winter night sky, is best observed from November to February.
Earlier this year, OpenAI demonstrated at a promotional event in Japan that the next-generation GPT model could be 100 times more powerful than GPT-4 under the same resource conditions. The company's presentation also explicitly marked "GPT NEXT" for release in 2024. OpenAI's goal is to achieve artificial general intelligence by integrating a series of large models.
Currently, OpenAI is at a critical stage of development. The company has just completed a $6.6 billion financing round, valuing it at $150 billion, and is undergoing a restructuring towards profitability. With the departure of former Chief Technology Officer Mira Murati, only Altman remains from the core management team of the GPT-4 era. Under such circumstances, the progress of the new model's development is highly anticipated.