OpenAI Unveils GPT-4o Model with Real-time Visual Inference, Available for Free to All
Yesterday, OpenAI held a roughly 30-minute spring update release conference, which surprisingly didn't introduce a search engine, but instead launched the GPT-4o model, based on GPT-4, The model features text, speech, and visual real-time reasoning.
This model offers GPT-4 level abilities but with faster speeds, allowing users to input text, voice, or upload images for inference. Moreover, GPT-4o can also turn on the camera to analyze the screen content in real time. For a simple example: when you travel abroad, you can take GPT-4o to take pictures of the surrounding scenes and let ChatGPT help you translate road signs or provide various suggestions.
ChatGPT that calls the GPT-4o model can conduct smooth conversations in real time, with a delay of only 232 milliseconds, while the delay of GPT-3.5 is about 2.5 seconds, which means there is a relatively obvious "stuck", which is no longer the case with GPT-4o , coupled with the extremely outstanding reasoning capabilities of GPT-4o, voice assistants such as Siri seem like toys.
Earlier, Apple and OpenAI reached an agreement to integrate OpenAI's chatbot into iOS 18, which is likely to be driven by GPT-4o, providing users with enhanced natural language conversation capabilities.
What's unexpected is that GPT-4o will be available for free to all users. Currently, a limited number of ChatGPT users have access to GPT-4o without needing a ChatGPT Plus subscription.
Free users will have quota limitations, but subscribing to ChatGPT Plus will allow for more conversations. OpenAI will also roll out the GPT-4o model to enterprises later.
Faster than grayscale permissions is API permissions. Now all developers can get access to the GPT-4o model, but it is not free, but the rate is only half of the GPT-4 series and faster.
Additionally, OpenAI has released a ChatGPT for Mac client, currently in testing, with the installation file already circulating online.
Finally, due to GPT-4o's exceptional abilities, some real-time translation, learning, and training applications or services may face significant pressure, such as language learning app Duolingo, whose stock price has dropped, as investors consider the possibility of being replaced by such AI applications.
Interested users can visit OpenAI's official website for more information: https://openai.com/index/hello-gpt-4o/