Sam Altman from OpenAI discussed the upcoming models GPT-4.5 and GPT-5, emphasizing the need for a simplified, unified interface that intelligently selects models based on user prompts. GPT-5 will integrate various functionalities like voice, image generation, and web search into a single model, enhancing user experience while offering tiered access based on subscription levels.
In a recent update, Sam Altman from OpenAI discussed the upcoming models GPT-4.5 and GPT-5, highlighting the current confusion surrounding the model selection process for users. Many users, including those with pro accounts, find the multitude of options overwhelming, with various models and features like web search and deep research toggles complicating the experience. Altman expressed a desire for a simplified, unified interface that intelligently selects the appropriate model based on user prompts, rather than requiring users to navigate through numerous options.
Altman revealed that GPT-4.5, internally referred to as Orion, will be the last model that does not incorporate “thinking” capabilities. This marks a significant transition, as GPT-4.5 will serve as a bridge to more advanced models that can understand and process user intent more effectively. The goal is to unify the various models in the OpenAI ecosystem, allowing for a seamless experience where the model can determine whether extended thinking or specific tools are necessary based on the user’s request.
The discussion also touched on the integration of different technologies within GPT-5, which will replace the standalone O series models. GPT-5 is expected to provide a more cohesive experience by combining various functionalities, such as voice, image generation, and web search, into a single model. This integration aims to enhance the model’s ability to perform a wide range of tasks without requiring users to specify which tools to use.
For users, the updates include changes to the free tier of ChatGPT, which will now offer unlimited access to GPT-5 at a standard intelligence level, subject to certain usage thresholds. Plus and pro subscribers will have access to higher intelligence levels and additional features, similar to how Apple offers different hardware configurations for its laptops. This tiered approach aims to provide users with varying levels of service based on their subscription.
Finally, Altman shared updates on current usage capabilities, including support for file and image uploads in ChatGPT, which enhances the model’s functionality. Additionally, the limits for the O3 mini High model have been significantly increased for plus users. These developments indicate OpenAI’s commitment to improving user experience and intelligence across its platforms, with a focus on creating a unified experience across web, mobile, and desktop applications.