Summary of Key Points from the Conference Call Company Overview - OpenAI has officially released GPT-5.4, introducing a "Thinking" mode across its ChatGPT, API, and Codex platforms [1] Core Insights and Arguments - The new GPT-5.4 model enhances deep web research capabilities, improving context retention during prolonged reasoning tasks [1] - Users can now interrupt the model's responses to add instructions or adjust directions, enhancing interactivity [1] - The guiding feature has been launched on Android and web platforms, with an iOS version expected to follow soon [1] - OpenAI developers have indicated that GPT-5.4 possesses native computer usage capabilities [1] - Codex and API now support up to 1 million context tokens, allowing for optimal coding of complex tasks and providing scalable tool searches within a larger ecosystem [1] - The model is designed for more efficient reasoning in long-duration, tool-intensive workflows [1] Additional Important Content - The release of GPT-5.4 Pro is specifically optimized for complex tasks, indicating a focus on enhancing productivity and efficiency in various applications [1]
未知机构:OpenAI正式发布GPT54提供Thinking思维模式上线C-20260306
2026-03-06 02:20