OpenAI CEO Sam Altman posted on social media at 10:23 a.m. US time today, officially saying goodbye to the GPT-4 model. He wrote: "Goodbye, GPT-4. You started a revolution. We will proudly preserve your data on a special hard drive for future historians."
GPT-4 will be retired, and GPT-4o will become the main model for ChatGPT
According to OpenAI's announcement, from April 30, 2025, the ChatGPT interface will be fully upgraded to use the GPT-4o model, and GPT-4 will only be available in the API service.
OpenAI said that GPT-4o has surpassed GPT-4 in all aspects, including writing, programming, and STEM fields, especially in command understanding, logical reasoning, and conversational fluency. GPT-4o, the new generation model, is the natural successor to GPT-4.
What is the difference between GPT-4 and GPT-4o?
GPT-4 was released in March 2023 and is a major advancement after GPT-3.5. It not only enhances multi-stage reasoning and complex text understanding capabilities, but also introduces stronger security and accuracy for the first time.
GPT-4o ("o" stands for omni) was released on May 14, 2024. It is OpenAI's first truly multimodal chat model that can simultaneously process text, voice, and image inputs and generate output in any form (such as voice reply and image generation). In addition, GPT-4o can support interactions in 50 languages, which is a groundbreaking update in cross-language and cross-media processing.
For users, GPT-4o not only responds faster and understands semantics more naturally, but also allows interaction through voice and images, greatly expanding the application scenarios of ChatGPT as an AI assistant.