OpenAI Unveils 25 Major DevDay Launches: 7 Powerful AI Updates You Need to Know

OpenAI has unveiled a sweeping collection of new AI products, models, developer tools and ChatGPT features at its 2026 DevDay, marking one of the company’s biggest product announcements to date.

Held in San Francisco on September 29, OpenAI DevDay introduced more than 20 major announcements spanning AI agents, models, APIs, collaboration tools, Codex, ChatGPT and enterprise services.

Among the biggest announcements are Dots, persistent AI agents designed to work on ongoing goals, GPT-6.1 Sol, a lower-cost model with near-Astra performance, and ChatGPT Space, a shared environment where teams and AI agents can work together.

OpenAI DevDay 2026: The Biggest Launches

Here are the major announcements from the event.

1. Dots: Always-on AI agents

OpenAI introduced Dots as persistent AI agents that can take responsibility for ongoing tasks rather than simply responding to individual prompts.

A Dot has its own cloud computer and browser and can continue working between conversations. Users can give it a goal, connect supported applications and review its work as it progresses.

OpenAI is also testing specialist Dots designed for specific responsibilities inside organizations.

2. GPT-6.1 Sol

GPT-6.1 Sol is a new model designed to deliver near-Astra performance at a lower cost.

OpenAI says the model is aimed at complex coding, computer use and professional work. Its API pricing is listed at $2 per million input tokens and $10 per million output tokens.

3. Ultrafast mode

OpenAI also introduced Ultrafast, a paid speed tier designed to generate responses substantially faster.

The company’s developer documentation describes GPT-6 Astra Ultrafast as a faster-output option in the Responses API.

4. ChatGPT Space

ChatGPT Space gives teams a shared environment where people and AI agents can work with common files and project context.

The feature is designed around collaborative work, allowing teams to bring together colleagues, Pages, Dots and other AI-powered capabilities.

5. Pages

OpenAI introduced Pages as editable documents where people and agents can work together.

The feature is designed to make AI collaboration more practical for documents, charts and interactive work rather than keeping everything inside a traditional chat interface.

6. A new $500 Pro plan

OpenAI introduced a Pro 500 subscription priced at $500 per month.

The company says the plan provides a significantly larger usage allowance and access to Ultrafast capabilities.

OpenAI has also reopened subscriptions to the $200 Pro tier, although new subscriptions have different allowances from some existing grandfathered plans.

7. Agents API

The new Agents API gives developers hosted infrastructure for building and running AI agents.

It includes capabilities such as execution, memory, tools and multi-agent support. OpenAI also added computer-use capabilities that allow agents to operate software through its user interface.

8. Decisions API

OpenAI introduced the Decisions API for applications that need fast, consistent decisions from predefined choices.

The API can use Luna to classify inputs, route requests or select an action from a fixed set of possible answers.

9. Sign in with ChatGPT

Developers can now allow users to sign into other products using their ChatGPT account.

OpenAI says the feature launched with 16 partner applications, while eligible subscription usage can also be used through participating products.

10. ChatGPT plugins and extensions

OpenAI is expanding ChatGPT into a platform where third-party applications can operate through plugins and connected tools.

The company is also introducing MCP events, allowing connected applications to trigger automations.

11. OpenAI Marketplace

OpenAI has launched a marketplace for partner applications, particularly for eligible enterprise customers.

The marketplace allows qualifying organizations to use part of their OpenAI commitments toward partner products.

12. OpenAI agents inside AWS

Amazon Bedrock Managed Agents can now run OpenAI-powered agents inside AWS environments, expanding the company’s enterprise deployment options.

13. Private Intelligence

OpenAI announced Private Intelligence, including support for zero data retention and Private Safety Processing.

The company also said a Private Inference preview is planned for later this year.

14. Codex moves further into the cloud

Codex can now continue running cloud-based work without requiring developers to keep their computers open.

This allows coding tasks to continue in the background while developers move on to other work.

15. Voice for Codex CLI

OpenAI is adding voice capabilities to the Codex command-line workflow, allowing developers to interact with Codex using spoken instructions.

16. Faster computer use

OpenAI also highlighted improvements to computer-use capabilities, allowing AI agents to interact with software and websites more quickly.

17. Codex pull-request reviews

Codex can review pull requests directly within the development workflow, helping developers identify issues before code is merged.

18. Codex security

OpenAI is expanding Codex into security workflows, with capabilities designed to identify vulnerabilities and help developers address them.

19. Open-source Codex harness

OpenAI also announced the opening of its Codex harness, giving developers access to more of the infrastructure behind agentic coding workflows.

20. More capabilities for Sites

OpenAI is expanding Sites with databases, scheduled tasks and additional automation capabilities.

This moves Sites beyond simple AI-powered experiences toward applications capable of storing information and performing recurring work.

21. PowerPoint export

Users can now export presentations directly into PowerPoint, making it easier to move AI-generated slide work into established office workflows.

22. Specialist Dots for organizations

Organizations can create specialist Dots designed around particular responsibilities.

OpenAI says these agents can eventually be shared across teams, allowing companies to build AI workers around specific roles and recurring tasks.

23. ChatGPT in Slack and Microsoft Teams

ChatGPT can now be mentioned directly inside Slack and Microsoft Teams.

The assistant can respond to mentions and use connected tools within those conversations, bringing AI assistance into existing workplace communication channels.

24. Meetings plugin

OpenAI introduced a Meetings plugin that can generate meeting notes and action items.

The feature is currently in beta on macOS for Pro and Business users.

25. More collaborative work between people and AI

Taken together, the announcements show OpenAI pushing ChatGPT beyond a traditional chatbot model.

Dots can handle ongoing tasks, Spaces bring people and agents into shared environments, Pages support collaborative documents, and Codex continues moving deeper into software development workflows.

OpenAI’s developer platform is also becoming more open to third-party applications through plugins, MCP connections, the Marketplace and Sign in with ChatGPT.

What OpenAI’s DevDay announcements mean

The common thread across the announcements is a shift from AI that simply answers questions to AI that can perform and coordinate work.

Dots are designed to operate continuously. Agents API gives developers infrastructure for building similar systems. ChatGPT Space and Pages allow humans and AI agents to collaborate, while Codex increasingly handles longer-running software development tasks.

GPT-6.1 Sol also gives developers another option between capability and cost, with OpenAI positioning it as a near-Astra model for complex work at a lower price.

With these launches, OpenAI is expanding its AI ecosystem across consumer products, enterprise collaboration, developer APIs and autonomous agents.

The full impact will depend on availability, pricing, reliability and how developers and organizations use the new tools as they roll out.