OpenAI 首席执行官 Sam Altman 宣布对 ChatGPT 实施“code red”，要求公司内部资源转向加速改进核心产品，并延缓自主 AI agents、广告与购物等其他项目。措施包括临时团队调动、跨团队每日会议以及强化对模型行为、图像生成与可定制化体验的改进。内部消息称 OpenAI 正在研发代号为“Garlic”的新模型，在推理与编码任务上表现良好。此举反映大型 AI 开发商之间竞争升级，尤其是在 Google 于上月发布 Gemini 3 并获得快速赞誉之后。
市场竞争方面，Google 表示其 Gemini 应用的用户规模达 6.5 亿人，并在 2025 年 10 月实现 7,300 万月度下载量；Sensor Tower 的数据称 ChatGPT 则维持领先，达到 9,300 万月度下载量，并拥有 8 亿每周活跃用户。Alphabet 的 Gemini 3 强调推理、编码与处理以往令聊天机器人失误的专业任务，而 OpenAI 则在 2025 年 10 月推出首款 AI 驱动的网页浏览器，以正面迎战。与此同时，Anthropic 在上月释放新模型以强化自动化编码与办公任务功能，加剧企业用户争夺。
OpenAI 高层如 ChatGPT 负责人 Nick Turley 强调，公司现阶段重心是提升 ChatGPT 能力、扩大全球覆盖，并使交互更直观个性化。部分员工已被要求从广告与 AI agents 等方向撤离，转而支持 ChatGPT 相关开发。Altman 在内部 Slack 通知中表示，目标是实现用户体验的更深层定制化、改善模型行为并提升图像生成质量。整体趋势显示，各大参与者正在向更高性能、强推理与企业级应用场景加速靠拢。
OpenAI CEO Sam Altman has declared a “code red” to redirect internal resources toward accelerating improvements to ChatGPT, delaying work on autonomous AI agents, advertising, and shopping features. Measures include temporary team transfers, daily cross-team calls, and a focus on refining model behavior, image generation, and user-experience customization. Sources say OpenAI is also developing a new model code-named “Garlic,” which has performed strongly on reasoning and coding tasks. The escalation reflects intensifying competition among major AI developers, especially following Google’s release of Gemini 3 last month to rapid acclaim.
Competitive metrics show Google reporting 650 million users of its Gemini app and 73 million monthly downloads as of October 2025, while ChatGPT leads with 93 million monthly downloads and 800 million weekly users, according to Sensor Tower. Gemini 3 emphasizes reasoning, coding, and niche task performance, while OpenAI countered by launching its first AI-powered web browser in October. At the same time, Anthropic released a new model last month aimed at automating coding and office workflows, adding pressure in the battle for enterprise adoption.
OpenAI leaders such as ChatGPT head Nick Turley reiterated that the company’s top priority is to make ChatGPT more capable, globally accessible, and more intuitive and personalized. Employees are being shifted away from advertising and AI-agent work to reinforce ChatGPT development. Altman said in Slack that the goal is deeper customization, improved behavioral reliability, and stronger image-generation quality. Overall trends indicate that major players are accelerating toward higher-performance models emphasizing reasoning strength and enterprise-grade functionality.