Core Insights - OpenAI has announced the upgrade of its flagship AI model to GPT-5.1, which will start rolling out to ChatGPT users this week, enhancing the model's intelligence and engagement compared to the previous version released in August [1][3] Group 1: Model Features - GPT-5.1 is divided into two versions: "Instant" and "Thinking," with the former excelling in temperature, intelligence, and instruction comprehension, while the latter is more efficient in handling complex problems and faster in simple tasks [3] - Users can now switch between various personality presets such as "default, professional, friendly, candid, quirky, efficient, nerdy, and cynical," allowing for a more tailored interaction experience [3] - The original GPT-5 model will remain available in ChatGPT as a "legacy model" for three months before being phased out [3] Group 2: User Base and Market Response - ChatGPT has surpassed 800 million users, indicating a demand that exceeds what a standard model can handle [3] - The initial release of GPT-5 did not meet user expectations, leading to dissatisfaction and the reintroduction of GPT-4o shortly after its launch [3] Group 3: Strategic Partnerships - Following the underwhelming performance of GPT-5, OpenAI's strategic partner Microsoft is exploring collaborations with other AI companies like Anthropic, integrating related models into various products such as Copilot Researcher and Office Agent [4] - OpenAI recently launched the AI-driven web browser ChatGPT Atlas, currently available only to premium subscribers, which allows the model to perform tasks directly in the browser [4]
OpenAI发布GPT-5.1模型 更具“温暖感” 并提供更多个性化选项
Feng Huang Wang·2025-11-14 09:05