Tech Desk 
Feb 13, 2025
OpenAI is gearing up to launch GPT-4.5 and GPT-5 to offer major improvements over ChatGPT as the company's CEO Sam Ataman shared a roadmap for these AI models, hinting at their release in the coming weeks or months.
This announcement comes as DeepSeek, a powerful and affordable Chinese AI model, gained global attention. OpenAI has even suggested that DeepSeek may have been trained using OpenAI’s models, adding more competition to the AI industry.
OpenAI set to launch two new AI models
One of the biggest changes OpenAI is planning to introduce is to simplify its AI offerings. Currently, users have to choose between models like GPT-40, GPT-4 Turbo, o1, and o1 mini, all with varying levels of reasoning and speed.
Moreover, Altaman confirmed that OpenAI will unify these models so that ChatGPT automatically picks the best AI model for each task. This will remove the need for users to manually select a model, making the AI experience smoother.
Before this transition, OpenAI will first introduce GPT-4.5, known internally as “Orion.” Altman described it as OpenAI’s last non-chain-of-thought model, meaning it won’t focus on deep, step-by-step reasoning like GPT-5 will.
Once GPT-5 is launched, ChatGPT will be able to use all available tools and adjust its reasoning pattern based on the task at hand.
Furthermore, a major highlight is that GPT-5 will be available to free users at a standard intelligence level, while ChatGPT Plus and Pro subscribers will get features like voice, deep research, and a smarter AI experience.
