OpenAI GPT-4o
Who makes it?
OpenAI, a leading AI research and deployment company, created GPT-4o. The company is known for its innovative AI models and products, including its popular AI chatbot ChatGPT.
What is it?
GPT-4o is a new iteration of the GPT-4 model. According to OpenAI (and my initial tests with it), it’s faster and more capable when inputting multiple forms of data. It improves ChatGPT’s capabilities by offering a much faster response time for inputs, especially spoken audio, and improved performance in handling multimodal inputs, including text, speech, and images. This multimodal approach gives it the GPT-4o name, with the “o” standing for “omni.”
CNET has compiled a great two-minute supercut video showcasing some of the newest features of GPT-4o if you want a visual for what these updates look like.
When/Where is it available?
GPT-4o’s capabilities are rolling out iteratively, with text and image functionalities starting May 13th, 2024. The full rollout, which will include the new ChatGPT voice experience, will happen across OpenAI’s developer (OpenAI’s API and Microsoft’s Azure OpenAI Service) and consumer-facing products (ChatGPT) over the next few weeks, according to OpenAI.
Why should I care?
GPT-4o stands to offer significant improvements over the previous GPT-3.5 Turbo and GPT-4 models. Its ability to handle text, speech, and images throughout the same input will make it a versatile tool for more capable applications, from real-time voice assistants (such as real-time translation of conversations in two languages) to advanced image analysis (such as detecting the emotion on a person’s face in real-time). OpenAI says this new model will expand the potential uses of AI in everyday tasks and professional environments even more so than its already impressive predecessors.
How much does it cost?
GPT-4o is free for all ChatGPT users, but usage limits do apply. ChatGPT Plus and Team users will still have limits for the time being, but they’ll be up to “5x higher” than those imposed on the free users, according to OpenAI.