IT之家 9 月 30 日消息，在北京时间 9 月 30 日凌晨 1 点举行的 OpenAI DevDay 2026 活动中，OpenAI 发布了 dots。
dots 是一个全天候在线的智能体，使用四个可爱的卡通形象，可以协助使用者规划议程、外观网站、测试应用、修改笔记等。
另外，它会随着时间的推移逐步了解使用者的工作习惯，以及使用者想要将注意力集中在哪些维度。它可以查看夜间收到的信息，并在早晨开始一天的工作之前，将所有紧急事项及时通知使用者。
dots 还可以帮使用者处理生活中的小事，例如订机票、晚餐时间，或者使用者一直拖着没做的家装；使用者也可以和 dots 语音打电话聊天。
依据 OpenAI 官方说法，dots 还可与 ChatGPT 平台上的 4,000 多款应用协同工作。当使用者将设备接入平台时，它将可以使用已连接的相同插件；dots 具备独立的云端电脑，能在长期使用中从反馈里学习，并全天候为达成使用者的目标而工作。
dots 可以接手项目并自主推进，即使它同时还在处理其他几个项目。使用者可以不断向它提出新任务和新想法，无需在不同对话间来回切换，也不必逐步指导。
无论借助 ChatGPT、短信、Slack 还是 Teams，使用者都能随时联系到 dots。使用者可以提问、探索想法、引导工作方向；需要聊一聊、理清思路时，也可以与它开展语音通话。合作越多，dots 就越了解你的偏好、思考方式，以及你心目中好工作的标准。
IT之家从大会获悉，dots 由 OpenAI 的 GPT‑6 Astra 模型给出兼容，并内置了多重保护与安全机制。使用者可以设置应用和电脑的使用时限，并且可以为智能设备执行的任何操作自定义指令。
每个 dot 都在独立的云端电脑上工作。除非使用者挑选连接，否则电脑及其中的内容会与它保持隔离。对于受兼容的登录方式，dots 可以使用已保存的密码，而不向模型暴露密码。
即使使用者没有直接与 dots 协作，它也会始终在后台寻找能帮上忙的事情。OpenAI 将其称为“主动研究”。它借助权限受限的工具使用已连接的应用来开展研究，因而无法发送消息、修改应用内容，也无法控制浏览器或电脑。
使用者可以挑选 dots 能访问哪些应用，并借助现有的 ChatGPT 应用控制设置管理其权限。dots 从一开始就有内置规则，明确何时可以独立操作，何时需要请求批准。借助自定义规则，使用者可以允许特定操作、要求事先批准，或禁止这些操作。
即日起，dots 将在 ChatGPT 中向符合条件的市面内的 Pro 和 Business Premium 使用者逐步开放。工作空间管理员启用后，Enterprise 使用者即可试用测试版。
Pro 或 Business Premium 套餐已包含首个 dot，无需额外付费。套餐还包含处理更深入工作的额度，发布后的首个月将给出更高限额。
将来，使用者将可以按月支付固定费用，添加更多 dot，或提高 dots 的工作速度及可承担的工作量。
与 dots 的对话不会计入 ChatGPT 使用限额。当使用者让 dots 在 Codex 或“ChatGPT 工作”中启动或管理任务时，这些任务仍会照常计入使用限额。
本文来源：https://www.ithome.com/1/008/523.htm
若有版权问题，请来信告知：13951431913#139.com 或在右方提交评论。
Thanks to IT之家 netizens 我家小妞很拽, HuTao29, simonsmh, and Nuc_F for their tip submissions!
IT之家 news on September 30: At the OpenAI DevDay 2026 event held at 1:00 AM Beijing Time on September 30, OpenAI released dots.
dots is an always-online agent that uses four adorable cartoon avatars to help users plan agendas, visit websites, test applications, edit notes, and more.
Additionally, it gradually learns about users' work habits over time and which dimensions users want to focus their attention on. It can review information received overnight and notify users of all urgent matters in a timely manner before they start their day in the morning.
dots can also help users handle small matters in daily life, such as booking flights, dinner plans, or home improvement tasks users have been putting off; users can also chat with dots over voice calls.
According to OpenAI's official statement, dots can also work collaboratively with more than 4,000 applications on the ChatGPT platform. When users connect their devices to the platform, it can use the same connected plugins; dots has an independent cloud computer, can learn from feedback during long-term use, and works around the clock to achieve users' goals.
dots can take over projects and advance them autonomously, even while it is simultaneously handling several other projects. Users can continuously give it new tasks and ideas without needing to switch back and forth between different conversations or provide step-by-step guidance.
Whether through ChatGPT, SMS, Slack, or Teams, users can reach dots at any time. Users can ask questions, explore ideas, and steer the direction of work; when they need to talk things through or sort out their thoughts, they can also have voice calls with it. The more you collaborate, the better dots understands your preferences, your way of thinking, and your standards for good work.
IT之家 learned from the conference that dots is powered by OpenAI's GPT-6 Astra model and comes with multiple built-in protection and security mechanisms. Users can set time limits for app and computer usage, and can customize instructions for any action the intelligent device performs.
Each dot works on an independent cloud computer. Unless the user chooses to connect, the computer and its contents remain isolated from it. For compatible login methods, dots can use saved passwords without exposing the passwords to the model.
Even when users are not directly collaborating with dots, it is always working in the background to find ways to help. OpenAI calls this "proactive research." It uses connected applications through permission-limited tools to conduct research, so it cannot send messages, modify app content, or control browsers or computers.
Users can choose which applications dots can access and manage its permissions through the existing ChatGPT app control settings. dots has built-in rules from the start that clarify when it can operate independently and when it needs to request approval. With custom rules, users can allow specific actions, require prior approval, or prohibit these operations.
Starting today, dots will gradually roll out to eligible Pro and Business Premium users in ChatGPT. After workspace administrators enable it, Enterprise users can try the beta version.
The Pro or Business Premium plan already includes the first dot at no additional cost. The plan also includes credits for handling deeper work, with higher limits available in the first month after release.
In the future, users will be able to pay a fixed monthly fee to add more dots, or to increase dots' work speed and the workload they can handle.
Conversations with dots do not count toward ChatGPT usage limits. When users ask dots to start or manage tasks in Codex or "ChatGPT Work", those tasks will still count toward usage limits as usual.