Meet Dots: Your 24/7 AI Co-pilot
OpenAI unveiled Dots, a new proactive, always-on personal AI agent, at its DevDay 2026 conference on September 29. Designed to operate 24 hours a day, Dots anticipates user needs rather than merely responding to direct prompts. The agent, powered by OpenAI's GPT-6 Astra model, operates continuously in the background, aiming to streamline daily tasks.
Dots enters a rapidly expanding market, positioning itself against rivals such as SpaceXAI's Grok Bot and Meta's Muse. This new generation of highly personal AI assistants draws significant inspiration from Peter Steinberger's OpenClaw, which established the foundational concept. Industry observers note a converging design language, with many agents, including Dots and Grok Bot, adopting soft geometric shapes with integrated eyeballs.
Dots integrates directly within the ChatGPT application, a contrasting approach to some standalone competitors. It connects to personal accounts, including:
- Gmail
- Google Docs
- Calendars
This integration allows Dots to access and manage personal data from its dedicated cloud environment. While conversations with Dots do not count towards standard ChatGPT usage, controlling ChatGPT or Codex threads does. Dots is initially available to subscribers of ChatGPT's Pro and Business Premium plans. Matthew Berman, an expert who attended DevDay, expressed a preference for a standalone application, citing concerns about computer control capabilities within an existing app.
AI Gets Cheaper... And Way More Expensive
OpenAI's DevDay 2026 introduced GPT-6.1 Sol, a highly efficient new model. Sol delivers performance nearly on par with the flagship GPT-6 Astra, but at a fraction of the cost. Its pricing stands at $2 per million input tokens and $10 per million output tokens, significantly cheaper than Astra's $10 input and $50 output rates. Sol also offers a cached million input tokens for just 10 cents.
The event also highlighted a premium tier for speed, with the new Ultrafast mode for GPT-6 Astra. Powered by Cerebrus chips, Ultrafast boosts Astra's speed by 8x compared to its normal operation on Nvidia GPUs. However, this cutting-edge velocity comes at a severe cost: users pay 6x more, with tokens burning rapidly.
OpenAI introduced the Pro 500 plan, priced at $500 per month. This plan offers 25 times the usage limits of the Plus plan, which itself was reduced from 20x to 10x usage. This tiered pricing strategy indicates a market segmentation where core AI intelligence is becoming more commoditized and accessible, while the absolute frontier of speed and capability commands a substantial premium.
Codex Unleashed: The Cloud is Your New IDE
OpenAI announced Codex will move natively to the cloud, eliminating bottlenecks previously imposed by local developer machines. This strategic shift enables true cloud-based development, allowing developers to code and deploy directly within OpenAI’s infrastructure. The move addresses the increasing performance gap where local processing, particularly for tool calling and terminal commands, now lags behind advanced inference speeds like those offered by Ultrafast mode.
Developers gained access to a suite of new tools. Codex Security Cloud launched, providing continuous, AI-powered vulnerability scanning for codebases, designed to identify and flag issues proactively. OpenAI also previewed the Decisions API, a fast, lightweight model engineered for rapid, efficient choice-making within applications, offering a streamlined approach to integrating quick decision logic.
A revamped plugin ecosystem was unveiled, signaling a major shift in the platform economy. Applications are now native to ChatGPT, streamlining user access and integration. Furthermore, a critical update allows users to "bring their own tokens" when interacting with third-party services via ChatGPT, granting greater control over resource consumption for external integrations.
Enjoying this? Get one like it in your inbox each morning.
one email a day · unsubscribe in two clicks · no third-party tracking
Beyond Chat: The New AI Workspace
OpenAI unveiled ChatGPT Space, a new collaborative platform designed to compete directly with existing workspace solutions like Notion. This native environment serves as a central hub for human teams and AI agents, enabling proactive, integrated workflows. Space aims to transform how teams interact with AI, moving beyond simple prompts to persistent, collaborative digital work.
Space integrates a wide array of digital assets, from internal documents and spreadsheets to external websites and applications. Autonomous AI agents, including the newly introduced Dots, can exist and operate directly within this shared workspace. Agents perform complex tasks, analyze information, and contribute to projects, eliminating the need for constant human oversight or manual data transfers between applications.
This launch represents a strategic expansion of OpenAI's influence, shifting AI from a conversational utility to an integrated operational backbone. The broader vision includes partnerships like the one with ModRetro, where Codex facilitates game development on the ModRetro Chromatic handheld hardware. This move positions OpenAI's technology as a foundational element for creation, collaboration, and automated decision-making across an expanding digital ecosystem.
Frequently Asked Questions
What is OpenAI Dots?
Dots is OpenAI's new personal AI assistant. It's designed to be 'always-on,' proactively managing tasks by connecting to your personal accounts like email and calendars, aiming to anticipate your needs before you ask.
What's the difference between GPT-6 Astra and GPT-6.1 Sol?
GPT-6 Astra is OpenAI's frontier model, offering maximum capability at a premium price ($10 input / $50 output per million tokens). GPT-6.1 Sol is a smaller, highly-optimized model that offers nearly the same performance as Astra but is significantly cheaper and faster ($2 input / $10 output per million tokens).
What is ChatGPT Space?
ChatGPT Space is a new collaborative platform from OpenAI, similar to Notion. It's a native environment where teams can work alongside AI agents on documents, spreadsheets, presentations, and more.
Is frontier AI getting more expensive?
Yes and no. The absolute frontier of performance, like the new 'Ultrafast' mode, is becoming more expensive with premium plans like the $500/month Pro 500. However, near-frontier intelligence, like that of GPT-6.1 Sol, is becoming dramatically cheaper and more accessible.