Codex负责人打脸Cursor CEO“规范驱动开发论”，18天造Sora爆款，靠智能体24小时不停跑，曝OpenAI狂飙内幕

Core Insights - Codex has experienced explosive growth since the release of GPT-5 in August, with a 20-fold increase in users and processing trillions of tokens weekly, making it OpenAI's most popular coding assistant [1][13] - The success of Codex is attributed not only to the model's improvements but also to a better understanding of how to integrate the model, API, and framework effectively [1][18] - OpenAI's unique organizational culture emphasizes rapid iteration and feedback, allowing for quick adjustments based on real-world usage [3][8] Group 1: Codex's Performance and Features - Codex's long-duration task capability has been enhanced through a mechanism called "compression," allowing it to summarize learned information and continue tasks over extended periods [1][18] - The transition of Codex from a cloud-based model to a local IDE integration has made it more user-friendly, resulting in significant growth [2][15] - Codex is envisioned as a proactive team member in software development, participating in the entire process from planning to deployment [10][21] Group 2: Organizational Culture and Development Approach - OpenAI's approach is characterized by a "shoot first, aim later" philosophy, prioritizing product release and subsequent optimization based on user feedback [3][8] - The company has a strong emphasis on hiring top talent and fostering a bottom-up culture, which facilitates rapid development and innovation [3][8] - The organization recognizes that the limitations of AGI development are often human factors, such as input and review speeds, rather than model capabilities [3][8] Group 3: Future of AI and Coding Assistants - The future of AI assistants is expected to shift from passive tools to active collaborators, capable of understanding context and taking initiative in workflows [21][22] - OpenAI aims to create a system where AI can provide assistance without explicit user commands, enhancing productivity and collaboration [12][21] - The integration of AI in coding is seen as a way to enhance human capabilities rather than replace them, with engineers becoming more valuable as they collaborate with AI [30][31]