OpenAI prepares revolutionary GPT-5 instead of o3

OpenAI announced a radical change in its product strategy, canceling the planned release of the o3 model in favor of creating a unified next-generation solution. Company CEO Sam Altman announced the release of GPT-5 model, which will combine all of OpenAI’s key technologies, including o3 developments.

The decision came as unexpected, considering that just a few weeks ago OpenAI’s Product Director Kevin Weil confirmed plans to launch o3 in «February-March». Initially, the model’s development was announced in December with a planned release in early 2025.

«We want to better inform about our plans and significantly simplify our product offerings,» Altman wrote in a post on X. «We want AI to ‘just work’ for you. We understand how complex our models and products have become. We, like you, don’t like model choice [in ChatGPT] and want to return to the magic of unified intelligence.»

The company also announced a new access structure for GPT-5:
— Basic users will get unlimited chat access at «standard intelligence level»
— ChatGPT Plus subscribers will be able to use «enhanced intelligence level»
— ChatGPT Pro users will get access to «even higher intelligence level»

Before launching GPT-5, the company plans to release an intermediate model GPT-4.5 (codename «Orion») in the coming weeks. According to Altman, this will be the last model without a «chain of reasoning,» which is typically less reliable in areas like mathematics and physics.

The new GPT-5 model will integrate all existing features, including voice interface, Canvas, search, and in-depth research. The company’s main goal is to create a universal system capable of effectively using all tools and adapting to a wide range of tasks.