OpenAI has decided to cancel the launch of its anticipated o3 AI model and instead focus on developing a more integrated and powerful model, GPT-5. CEO Sam Altman announced this change, stating that the company aims to simplify its product lineup and provide a more seamless AI experience. Rather than releasing separate models, OpenAI plans to create a unified system that can handle multiple tasks efficiently. GPT-5 will combine various technologies, including those initially intended for o3, into a single model for ChatGPT and API users.
Altman acknowledged that OpenAI’s current AI offerings have become complex and that the model selection process in ChatGPT needs improvement. The goal is to provide an AI that “just works” without users having to choose between different versions. He also mentioned that GPT-5 will be accessible to users with unlimited chat access at a standard intelligence level. At the same time, ChatGPT Plus and Pro subscribers will gain access to higher intelligence levels.
GPT-5 is expected to bring significant advancements, incorporating features like voice interaction, enhanced search, and deep research capabilities. It aims to create a more seamless AI experience by knowing when to take more time to process complex tasks.
Before launching GPT-5, OpenAI plans to roll out GPT-4.5, also known as Orion, in the coming weeks. This model will be the company’s last non-chain-of-thought model, meaning it may not be as reliable in reasoning-heavy tasks like math and physics. However, reasoning models, which fact-check themselves, tend to be more accurate.
Altman also acknowledged that Chinese AI lab DeepSeek’s recent R1 model has reduced OpenAI’s lead in AI development. In response, OpenAI plans to accelerate its release schedule to stay ahead in the competitive landscape.
However, Orion has reportedly faced performance challenges and may not be as groundbreaking as previous model upgrades.
