OpenAI has announced a broad expansion of its AI product lineup, introducing Dots, a new class of always-on personal AI agents, the GPT-6.1 Sol model, new ChatGPT collaboration capabilities, an upgraded Codex Security Cloud, and a $500-per-month ChatGPT Pro 500 tier that includes Astra Ultrafast.
The announcements significantly expand ChatGPT beyond a conventional conversational assistant into AI systems that can autonomously complete longer-running projects, operate computers, connect with enterprise applications, collaborate with teams, and work proactively in the background.
At the center of the announcement is Dots, which OpenAI describes as always-on AI agents powered by GPT-6 Astra.
Each Dot has its own cloud computer and browser, learns from feedback over time, and can work toward a user’s goals continuously rather than requiring a new prompt for every individual task.
Dots can connect with more than 4,000 applications through OpenAI’s plugin ecosystem, allowing them to interact with the software and information users already rely on.
Users can interact with a Dot through ChatGPT, Slack, and Microsoft Teams, including by messaging it or having a voice conversation. OpenAI also plans to add texting.
Unlike a conventional chat thread, a Dot can maintain context across channels and keep working even when the user isn’t actively interacting with it.
OpenAI said a Dot can simultaneously manage several projects without requiring the user to manually direct each step. It can use its own cloud computer for most tasks, while users can optionally give it permission to connect to and operate their laptop.
Potential applications span software development, research, sales, marketing, and content production.
For example, OpenAI said a Dot could monitor customer feedback, identify recurring software problems, build and test fixes, and prepare pull requests for a developer to review.
For researchers, a Dot could rerun analyses as new data becomes available, investigate unexpected results, update figures, and flag changes requiring review.
For content creators, it could process interview transcripts, identify clips, prepare show notes, and draft social posts based on a user’s established style and preferences.
OpenAI is also giving Dots the ability to perform “proactive research” when a user is not actively working with them.
During proactive research, the agent can inspect connected applications using restricted read-only tools. These tools cannot send messages, alter application content, or control the user’s browser or computer.
Users can specify which apps a Dot can access and create Custom Rules determining which actions it may perform autonomously, which require approval, and which are blocked.
OpenAI’s auto-review system evaluates actions that could affect accounts or disclose information and determines whether the action can proceed, needs user approval, or must be completed directly by the user.
Each Dot’s cloud environment remains separate from the user’s computer unless the user explicitly connects the two.
For supported websites, Dots can also use saved passwords without exposing those passwords directly to the AI model.
OpenAI said Dots are beginning to roll out to Pro and Business Premium users in eligible markets, while Enterprise customers, including eligible Edu and Healthcare organizations, can access the beta when enabled by an administrator.
The company is also testing specialist Dots designed to take on defined organizational responsibilities with their own identities, credentials, hardware, and access to enterprise systems.
Early internal applications at OpenAI have included procurement, invoice processing, email marketing, customer support, and commercial contracting.
OpenAI is also working with Microsoft to integrate specialist Dots with Microsoft Agent 365 governance and security controls.
Alongside Dots, OpenAI introduced GPT-6.1 Sol, an upgraded version of GPT-6 Sol intended to offer performance approaching GPT-6 Astra across agentic coding, professional work, and computer-use tasks at substantially lower cost.
OpenAI said GPT-6.1 Sol operates at approximately one-fifth of GPT-6 Astra’s standard input and output token prices.
API pricing is $2 per million input tokens, $0.10 per million cached input tokens, and $10 per million output tokens. The cached input price is 95% lower than GPT-6.1 Sol’s standard input price and 50% below GPT-6 Sol’s cached input price.
GPT-6.1 Sol is available in ChatGPT Work and Codex for Plus, Pro, Business, Enterprise, and Edu users, although it is not yet available directly in Chat.
Developers can access the model through the API as gpt-6.1-sol.
OpenAI also plans to introduce a GPT-6.1 Sol Ultrafast option in Codex that can generate tokens up to eight times faster than its standard speed.
The model showed significant improvements across OpenAI’s testing.
On DeepSWE v1.1, which evaluates long-horizon software engineering work in real codebases, GPT-6.1 Sol matched GPT-6 Astra at roughly one-fifth the cost and exceeded GPT-6 Sol’s best result by 6.4 percentage points.
On AutomationBench, GPT-6.1 Sol scored 2.2 percentage points above Opus 5.5 at medium reasoning effort while costing roughly one-third as much, and its score improved 4.8 percentage points over GPT-6 Sol at the same setting.
Computer-use performance also improved.
On OSWorld 2.0, GPT-6.1 Sol outperformed GPT-6 Sol by seven percentage points at maximum reasoning effort while costing less than half as much. Its result came within 2.1 percentage points of GPT-6 Astra while costing roughly one-seventh as much per task.
On Terminal-Bench Science 0.1, GPT-6.1 Sol more than doubled GPT-6 Sol’s score at maximum reasoning effort. Its average cost was $5.47 per task, compared with $23.21 for Opus 5.5 and $23.80 for GPT-6 Astra. Astra, however, remained the highest-scoring model in that scientific evaluation at 68.1%.
OpenAI also reported improved factuality.
At low reasoning effort, the percentage of responses containing at least one factual error in one deliberately difficult evaluation declined from 11.4% with GPT-6 Sol to 7.7% with GPT-6.1 Sol, representing a reduction of approximately 32%.
The product expansion is accompanied by a significant change to ChatGPT Pro pricing and tiers.
According to an OpenAI Help Center update published September 29, ChatGPT Pro now includes three monthly options: Pro 100 for $100 per month, Pro 200 for $200 per month, and a new Pro 500 plan for $500 per month.
Pro 500 provides the highest included usage and is the only Pro tier that includes Astra Ultrafast. Pro 100 and Pro 200 do not include Ultrafast.
OpenAI has also reopened Pro 200 to new subscriptions.
However, new subscribers who are not eligible for grandfathering receive a lower included usage allowance than the previous version of Pro 200, while the monthly price remains $200.
Eligible existing Pro 200 subscribers can retain their previous included usage allowance through October 29, 2026, after which their accounts will move to the updated allowance while remaining priced at $200 per month.
Among the Pro plans, Astra Ultrafast is available exclusively through Pro 500 at launch.
OpenAI said users who exhaust included usage for certain eligible features can purchase credits, with those credits supporting eligible use across products including Codex, ChatGPT Work, ChatGPT for Word, ChatGPT for Excel, and ChatGPT for PowerPoint.
OpenAI also introduced collaboration tools intended to make ChatGPT a more persistent workspace for teams.
ChatGPT Spaces will allow coworkers to organize uploaded files, AI-generated content, and shared information inside a common project environment that can be accessed by team members, ChatGPT, and Dots.
Spaces are expected to be available to Pro, Business, and Enterprise users, initially through ChatGPT’s desktop and web applications before expanding to mobile.
A new meetings plugin can capture meetings, generate notes and action items, and place the resulting information into a Space.
The captured audio is deleted once the notes are generated.
The meeting capability is initially being offered through the ChatGPT desktop application on macOS for Pro and Business customers, with Enterprise availability planned later.
ChatGPT is also being integrated more directly into Slack and Microsoft Teams for Business and Enterprise customers.
Users will be able to mention @ChatGPT inside channels, direct messages, and threads and ask it to perform tasks such as turning a conversation into presentation slides.
With administrator or user permission, ChatGPT can also access tools connected to Slack or Teams.
Another new capability, Pages, provides a collaborative document environment where users can generate text, charts, images, and visualizations and then share the work with colleagues for review.
OpenAI also plans to introduce collaborative presentation creation with exports to Microsoft PowerPoint and Google Slides, along with shareable profiles where users can showcase web applications and plugins.
The company is simultaneously strengthening its software security products.
Codex Security Cloud is being upgraded with access to OpenAI’s cyber-focused Daybreak Blue models included by default.
The system can scan entire connected GitHub repositories, continuously evaluate new commits, investigate and deduplicate potential vulnerabilities, and prepare fixes for human review even when a developer’s laptop is offline.
Codex Security Cloud is designed for managed, continuous security analysis of connected GitHub repositories, while Daybreak Blue is focused on defensive cybersecurity activities such as vulnerability discovery, validation, remediation, threat modeling, and security investigations.
The upgraded security capabilities are also available through a plugin in Codex desktop and web.
The announcements represent a substantial expansion of OpenAI’s strategy from AI that primarily responds to individual prompts toward AI that can maintain context, collaborate across applications, operate computers, execute multi-step workflows, monitor ongoing work, and take approved actions over extended periods.
Dots most clearly expresses that direction, while GPT-6.1 Sol lowers the cost of the underlying intelligence needed to run increasingly capable agents. ChatGPT Spaces and workplace integrations provide the collaborative environment for those agents, while Codex Security Cloud and OpenAI’s expanded permission systems are intended to provide additional oversight as AI systems gain more autonomy.
