OpenAI is preparing to launch the image understanding feature of GPT-4, named 'GPT-Vision', to compete with Google's Gemini. GPT-Vision will enable GPT-4 to have broader image application capabilities, generating text related to the image content. OpenAI is also developing a multimodal AI model named 'Gobi', which could become GPT-5. OpenAI plans to announce G at the developer conference on November 6.

OpenAI is preparing to launch the multimodal AI system GPT-Vision to compete with Google Gemini. OpenAI is also developing a more powerful multimodal model codenamed Gobi. The multimodal AI is entering a phase of practical application, leading to direct competition between OpenAI and Google.