OpenAI has decided to cancel its planned o3 AI model in favor of a more streamlined approach, integrating its latest advancements into a unified GPT-5 model. CEO Sam Altman announced that GPT-5 will incorporate various OpenAI technologies, including features originally planned for o3, to simplify AI offerings.
OpenAI initially planned to launch o3 early this year, but the company has now shifted its focus to making AI more intuitive and user-friendly. Altman acknowledged that OpenAI’s product range had become complex, leading to the decision to eliminate the standalone o3 model. Instead, GPT-5 aims to provide a seamless experience by combining multiple AI tools, including voice, canvas, search, and deep research.
Altman also revealed that once GPT-5 is released, it will be available with unlimited chat access at a standard intelligence setting, subject to specific usage limits. Users subscribed to ChatGPT Plus and ChatGPT Pro will gain access to more advanced intelligence levels within GPT-5.
OpenAI is also preparing to launch GPT-4.5, known as “Orion,” before GPT-5, marking its last non-reasoning model. Unlike reasoning models, which self-check their outputs to enhance reliability, non-reasoning models are less consistent in fields like mathematics and physics.
The decision to refine its AI roadmap comes amid increased competition from other AI developers, particularly China’s DeepSeek lab, which recently gained attention with its R1 model. DeepSeek’s advancements have challenged OpenAI’s dominance, prompting Altman to accelerate the company’s release schedule.
Reports indicate that Orion has faced technical setbacks, delivering a smaller performance boost over its predecessor, GPT-4o, compared to previous model upgrades. Despite these challenges, OpenAI remains focused on refining its AI technology to maintain its leadership in the industry.
