OpenAI has unveiled GPT-4o, an enhanced version of its GPT-4 language model that powers the popular ChatGPT assistant. According to Mira Murati, OpenAI’s Chief Technology Officer, the updated model offers significant performance improvements, boasting faster speeds and enhanced capabilities across text, vision, and audio modalities. While GPT-4o will be available free of charge to all users, paid subscribers will continue to enjoy up to five times higher capacity limits compared to free users.
In a blog post, OpenAI revealed that GPT-4o’s capabilities will be rolled out incrementally, with text and image functionalities becoming available in ChatGPT starting today. Sam Altman, OpenAI’s CEO, highlighted that the model is “natively multimodal,” enabling it to generate content and comprehend commands across various formats, including voice, text, and images. Developers interested in exploring GPT-4o will gain access to an API that promises to be twice as fast as GPT-4 Turbo while costing only half the price.
One of the notable new features introduced with GPT-4o is an enhanced voice mode for ChatGPT. This mode will allow the app to function as a real-time voice assistant, akin to the fictional AI system depicted in the movie “Her,” enabling it to observe its surroundings and respond accordingly. The current voice mode, in contrast, is more limited, responding to prompts one at a time and relying solely on audio input.
In a reflective blog post following the livestream event, Altman acknowledged a shift in OpenAI’s vision. While the company’s original goal was to “create all sorts of benefits for the world,” Altman recognized that the focus has evolved. Amid criticism for not open-sourcing its advanced AI models, Altman suggested that OpenAI’s role has shifted towards creating AI models and making them available to developers through paid APIs, allowing third parties to leverage these models to create innovative solutions that benefit society.
Prior to the GPT-4o launch, conflicting reports speculated that OpenAI might announce an AI search engine to rival Google, a voice assistant integrated into GPT-4 named Perplexity, or a entirely new and improved model called GPT-5. Notably, OpenAI timed this launch strategically ahead of Google I/O, the tech giant’s flagship conference, where various AI products from the Gemini team are expected to be unveiled.
