OpenAI’s DevDay 2026 was a big one. Rather than centering the event around a single new model, the company used its annual developer conference to expand nearly every part of its AI ecosystem, from coding and agents to ChatGPT plugins, collaborative workspaces, and new subscription tiers.

The company says DevDay 2026 brought more than 20 major announcements across ChatGPT, Codex, its models, and new ways for people and AI to work together. The broader theme was clear: OpenAI wants AI to move beyond the traditional chatbot window and become something that can work alongside people, carry out ongoing tasks, collaborate with teams, and operate across the software people already use.

Here’s everything OpenAI announced.

Dots are always-on AI agents

One of the biggest new products is Dots, which OpenAI describes as remarkably capable, always-on agents designed to handle ongoing responsibilities.

Unlike a conventional chatbot that waits for a prompt, Dots are intended to keep working on a user’s behalf. OpenAI says they can get to know what matters to you, take important work off your plate, and give you more time and attention for other things.

Dots are available on Pro and Business Premium in eligible markets. Enterprise, Education, and Healthcare customers can also try the beta when their workspace administrator enables it.

The launch represents a broader shift in how OpenAI sees AI assistants: not simply as tools you open when you need an answer, but as agents that can take responsibility for work over time.

GPT-6.1 Sol brings near-Astra intelligence to a cheaper model

OpenAI also introduced GPT-6.1 Sol, a major upgrade to GPT-6 Sol aimed particularly at agentic coding, computer use, and professional work.

The company says GPT-6.1 Sol delivers near-GPT-6 Astra intelligence while costing one-fifth of Astra’s standard input and output token prices. That lower cost is designed to give developers more room to run complex workflows without dramatically increasing their AI budgets.

GPT-6.1 Sol is available to API, Plus, Pro, Business, Enterprise, and Education users.

The positioning is interesting because OpenAI isn’t presenting Sol simply as a cheaper model. Instead, it is trying to bring much of the capability of its flagship intelligence into workloads where developers may need to run models repeatedly or for longer periods.

Ultrafast is OpenAI’s new speed tier

For applications where latency matters more than anything else, OpenAI introduced Ultrafast.

The premium speed tier can deliver up to 300 tokens per second in Codex, which OpenAI says is up to eight times faster, while API workloads can see up to six times faster token generation.

GPT-6 Astra Ultrafast is available today through the API and in ChatGPT Work and Codex for Pro 500 and Enterprise customers. GPT-6.1 Sol Ultrafast is coming soon.

That makes speed another axis on which OpenAI is now differentiating its models, alongside intelligence and cost.

Private Intelligence targets sensitive business workloads

OpenAI also announced Private Intelligence, a collection of technologies intended to make it easier for businesses to use advanced AI while keeping sensitive data protected.

One part is Zero Data Retention with Private Safety Processing, which allows automated safety reviews without giving OpenAI personnel access to the underlying content.

OpenAI is also previewing Private Inference, which is expected this fall. It combines confidential computing with strict, verifiable controls for businesses that need stronger assurances around how their data is handled.

Codex is moving into the cloud

Codex is getting a much broader home, too.

OpenAI says developers can now run Codex on a local computer, remotely from a phone, or in the cloud from essentially any device. Reusable development environments are designed to let tasks start quickly while giving teams a shared setup with approved settings and permissions.

Cloud Codex is available on Plus, Pro, Business, Healthcare, Education, and Enterprise plans.

The change makes Codex less dependent on the machine where development normally happens. A developer can start something on one device and continue working remotely without having to recreate the environment from scratch.

Codex CLI gets voice controls and multi-agent management

The Codex CLI is also getting a substantial refresh.

Developers can now start and steer tasks using their voice, while a new /agents view makes it easier to delegate work and keep track of multiple tasks.

OpenAI has also improved prompt editing, session resumption, worktrees, and the terminal interface for longer sessions.

The refreshed Codex CLI is available across all plans.

Code Review comes to the ChatGPT desktop app

OpenAI is bringing Codex deeper into the code-review workflow with a new Code Review experience in the ChatGPT desktop app.

Developers can review summaries, inspect diffs, and ask Codex about potential problems before submitting feedback on GitHub pull requests or GitLab merge requests.

There is also an automatic review option, allowing Codex to perform an initial pass in the cloud while the developer is away.

The feature is available on all plans.

Codex Security Cloud can continuously scan repositories

Security is another major focus.

Codex Security Cloud lets defenders scan entire GitHub repositories on demand or on a schedule. It can also continuously check new commits, investigate findings, remove duplicates, and prepare fixes in the cloud.

The system can continue working even when the developer’s laptop is closed.

Codex Security Cloud also includes access to models offered through Daybreak Blue without requiring a separate Daybreak application.

It is available to Pro, Business, Enterprise, and Education users on desktop and web.

Decisions API brings focused AI decision-making to developers

OpenAI introduced the Decisions API for applications that need AI to make decisions from a defined set of possible answers.

Developers can provide context using text or images and receive an answer based on user-defined questions with finite, predefined outcomes.

That makes the API suitable for tasks such as classification, routing requests, or deciding what an agent should do next.

It is currently in limited preview, with a broader release planned in the coming days.

Agents API gets computer use

OpenAI’s Agents API can now use computers, allowing developers to build agents capable of interacting directly with software to complete tasks.

The API also brings several capabilities from Codex, including multi-agent workflows, tool search, tool calling, and context compaction.

OpenAI manages the underlying infrastructure, allowing developers to focus on building their applications rather than operating the agent runtime themselves.

The computer-use capabilities are available through the API and in Codex and ChatGPT Work for Pro 500 and Enterprise customers.

OpenAI and AWS launch Bedrock Managed Agents

OpenAI has also teamed up with Amazon on Bedrock Managed Agents.

The service takes the core capabilities of OpenAI’s Agents API and adds AWS-native customization and integration with AWS resources.

The important distinction is that these OpenAI-powered agents can run entirely within AWS, giving organizations already operating there another way to deploy agentic applications.

ChatGPT is opening up with plugin extensions

OpenAI is opening the platform it uses internally to build ChatGPT features so developers can create their own experiences inside ChatGPT.

Plugin extensions can have a dedicated home in ChatGPT’s sidebar and provide interactive panels that work alongside conversations. Developers can also create viewers for supported file types.

The goal is to make ChatGPT more than a place where users interact with OpenAI-built features. Developers can now build experiences that live directly inside the ChatGPT interface.

Plugin extensions are available across all plans.

Building and discovering plugins is getting easier

OpenAI is also improving the plugin ecosystem itself.

Plugin Creator is designed to make building plugins easier, while a redesigned submission process gives developers clearer feedback.

OpenAI is also improving ranking and recommendations so users can more easily discover relevant plugins in the directory and inside conversations.

Users remain in control of which plugins they use and what access they approve.

Sites can now host ChatGPT plugins

OpenAI is extending plugins beyond ChatGPT itself.

Supported ChatGPT plugins can now be added to Sites that organizations build. Teammates can use the same application while keeping their own connected data and permissions.

OpenAI is also making automations easier to add and manage so Sites can keep shared information up to date.

The feature is available to Business, Enterprise, Healthcare, and Education customers.

MCP Events bring event-driven automations to plugins

Plugins are also gaining support for the proposed MCP Events specification.

That means a connected application can trigger an automation when something happens.

For example, a project-management system could notify ChatGPT when a new task arrives. ChatGPT could then read the relevant documents and prepare a plan without someone having to manually start the workflow.

OpenAI says MCP Events support is available across all plans.

ChatGPT Space creates a shared home for teams and AI

ChatGPT Space is a new collaborative environment where teammates, ChatGPT, and Dots can work from shared knowledge.

Teams can create a dedicated space for a project, while ChatGPT can help keep the space organized according to instructions provided by the team.

The feature is available on Pro, Business, and Enterprise plans on desktop and web. Finding, reading, and sharing pages are also available on mobile, while mobile creation and editing are coming later.

Pages turn ChatGPT into a collaborative document workspace

OpenAI is also introducing Pages, a new type of document designed specifically for collaboration between people and agents.

Pages can be used for writing, research, charts, image generation, and visualizing information. A page can be created from a conversation and shared with teammates for ideas and feedback.

Pages are available to Pro, Business, and Enterprise customers.

Collaborative slides are coming

OpenAI is bringing collaborative presentation creation to ChatGPT, too.

The upcoming feature will allow people and agents to create interactive slides from a conversation or an existing template.

Multiple teammates and agents will be able to edit a deck simultaneously and leave comments. Presentations can be shown inside ChatGPT or exported to PowerPoint and Google Slides while preserving formatting.

The feature is coming to Pro, Business, and Enterprise plans in the coming weeks.

Teams can now share tasks and recurring work

Business and Enterprise customers can now bring colleagues together in ChatGPT to share pages, slides, plugins, and spreadsheets.

Teams can also delegate recurring work, such as preparing weekly project updates.

ChatGPT can use connected tools to gather information and take action according to a schedule or in response to events such as a new email or Slack message.

Team members can refine the instructions together as their priorities change.

ChatGPT is coming directly into Slack and Microsoft Teams

OpenAI is making ChatGPT available inside the collaboration tools where many teams already spend their working day.

Users can mention @ChatGPT in a Slack or Microsoft Teams channel, thread, or direct message to get help with team work.

ChatGPT can use tools connected by an administrator or tools that the individual user has permission to access. Other teammates can then continue the same conversation and refine the result without requiring every participant to have an individual ChatGPT license.

The integration is available to Business and Enterprise customers.

Meetings plugin turns conversations into project follow-ups

The new Meetings plugin can take notes during meetings and save personalized summaries and action items in ChatGPT Space.

Those notes can remain private or be shared with a team. Users can then ask ChatGPT to turn the meeting output into a project plan or draft a follow-up.

OpenAI says meeting audio is deleted once the notes are ready and cannot be accessed or replayed.

The plugin is available in beta in the ChatGPT desktop app on macOS for Pro and Business users, with Enterprise support coming later.

Shareable profiles give developers a public showcase

OpenAI is also introducing Shareable profiles, which let people showcase the Sites and plugins they create so others can discover and reuse them.

For teams, profiles can also make shared skills easier to discover, although skills sharing remains limited to the workspace.

Shareable profiles are available to Business and Enterprise users.

Sign in with ChatGPT expands beyond OpenAI’s own products

OpenAI is also making it easier to take a ChatGPT subscription into other applications.

With Sign in with ChatGPT, users can use their ChatGPT plan with 16 participating partners, including Cognition’s Devin, Notion, Vercel, T3, OpenClaw, and Dactyl.

Users can control how much of their available usage each partner can consume, with eligible OpenAI usage counting toward the limits of their ChatGPT plan.

The identity component is available globally, while plan usage is currently available to Plus and Pro users in participating tools.

OpenAI launches a $500 Pro tier

For people who want substantially more usage and speed, OpenAI is introducing Pro 500.

The new tier offers 25 times the usage allowance of ChatGPT Plus and includes access to Ultrafast.

OpenAI says Pro 500 is available now.

This is also the plan that sits behind some of the company’s more demanding new features, including access to GPT-6 Astra Ultrafast in ChatGPT Work and Codex.

OpenAI Marketplace brings partner software to enterprise customers

Finally, OpenAI announced the OpenAI Marketplace for eligible enterprise customers.

Organizations can apply part of their existing OpenAI commitment toward approved partner software, giving them another way to buy and use software through their existing AI relationship.

The first 32 partners include Adobe and Figma for creative work; Sierra, Decagon, HubSpot, Salesforce, and ServiceNow for customer experience; Harvey and Legora for legal work; Palo Alto Networks and CrowdStrike for cybersecurity; and Baseten for open-source models.

Enterprise customers can now express interest in the marketplace.

DevDay 2026 is really about turning AI into infrastructure

Taken together, the announcements make DevDay 2026 feel less like a conventional model launch and more like a statement about where OpenAI wants its platform to go next.

GPT-6.1 Sol makes advanced intelligence cheaper to deploy. Ultrafast makes it faster. Agents API and computer use give developers ways to build systems that can actually perform tasks. Dots bring that concept directly to users.

At the same time, plugins, Spaces, Pages, collaborative slides, team tasks, Slack and Teams integrations, and Sign in with ChatGPT push OpenAI toward becoming a layer that sits across the software people already use.

That is probably the most important thread running through DevDay 2026. OpenAI isn’t just asking developers to build another chatbot. It is giving them more pieces for building AI that can operate continuously, collaborate with people, use software, and take responsibility for work.

And with more than 20 announcements arriving in a single day, there is a lot here for developers—and increasingly, ordinary ChatGPT users—to keep up with.