OpenAI DevDay 2026: The Agent Era Begins as Sam Altman Takes the Stage

OpenAI DevDay 2026 is happening today at Fort Mason in San Francisco, with CEO Sam Altman scheduled to deliver the opening keynote at 10:00 a.m. Pacific Time.

For developers, this year’s event comes at an important moment for the AI industry.

The market is moving beyond simply giving developers access to powerful foundation models through an API. AI companies are increasingly building complete agent products designed to perform tasks, operate across applications and work with less human intervention.

OpenAI now has an opportunity to show how it plans to compete in this rapidly changing market.

OpenAI’s official DevDay schedule confirms that the September 29 event will feature technical sessions, API and tool demonstrations, hands-on workshops and a keynote from Altman. The keynote will also be livestreamed for free.

OpenAI DevDay 2026 Is Taking Place Today

The event is being held at Fort Mason in San Francisco on September 29.

The official schedule lists:

• Breakfast at 8:00 a.m. PT
• Sam Altman’s opening keynote at 10:00 a.m. PT
• Developer breakouts beginning at 11:15 a.m.
• Closing session at 4:00 p.m.
• Reception at 4:45 p.m.

Developers who cannot attend in person can watch the keynote online.

For audiences in India, the keynote begins at 10:30 p.m. IST.

The Big Shift: From APIs to AI Agents

One of the biggest questions surrounding DevDay is whether OpenAI will place more emphasis on finished AI agents rather than simply providing increasingly capable models through APIs.

The shift is already visible across the technology industry.

AI companies are increasingly building systems that can:

• Browse the web
• Use software
• Manage files
• Complete multi-step tasks
• Communicate with other systems
• Run workflows autonomously

This represents a major change in how people interact with AI.

Instead of asking a model a question and receiving an answer, users increasingly want AI systems that can take responsibility for completing a task.

OpenAI’s GPT-6 Astra already introduced advanced computer-use capabilities, allowing the model to navigate websites, fill forms and interact with computer interfaces.

DevDay could reveal how OpenAI intends to turn those capabilities into more persistent products and developer platforms.

Is OpenAI Preparing an Always-On Agent Called “O”?

One of the most interesting rumours ahead of DevDay involves an apparent product referred to as “O.”

Reports have pointed to references to “O” inside ChatGPT configuration files and an apparent “-o” email suffix. The $100 ChatGPT Pro upgrade page has also reportedly contained references interpreted as an always-on assistant.

However, OpenAI has not officially confirmed that an “O” product is launching today.

If the reports are accurate, the concept would represent a significant evolution of the traditional chatbot.

An always-on agent could potentially continue working outside a normal chat session, manage tasks over time and interact with external services on a user’s behalf.

That would place OpenAI directly into the increasingly competitive market for persistent AI agents.

GPT-6 Cyber Could Be Another Major Announcement

Cybersecurity is another major area to watch.

Fortune reported last week that OpenAI is preparing GPT-6 Cyber, a cybersecurity-focused model designed to help organisations identify vulnerabilities and automate security workflows.

According to Fortune’s reporting, GPT-6 Cyber would be OpenAI’s fourth cybersecurity-focused model of 2026, following:

• GPT-5.4 Cyber
• GPT-5.5 Cyber
• GPT-5.6 Cyber
• GPT-6 Cyber

The report also says OpenAI is preparing a separate product designed to help customers deploy the model more securely and automatically.

A limited group of customers in OpenAI’s Daybreak Red cybersecurity programme has reportedly already been testing GPT-6 Cyber.

Importantly, Fortune’s original report said the model was expected to be previewed in the coming weeks, meaning its appearance at today’s DevDay should still be treated as an expectation rather than a confirmed announcement.

Why Cybersecurity Matters So Much Right Now

The timing of GPT-6 Cyber is particularly significant.

OpenAI and other frontier AI developers have faced increased scrutiny following incidents in which AI agents accessed external systems during testing.

GPT-6 Astra itself was released earlier this month with advanced computer-use and cybersecurity capabilities. OpenAI described Astra as its first model to reach the critical cybersecurity capability level under its own safety framework.

That creates an unusual situation.

The same technology that can potentially help defenders identify vulnerabilities can also provide increasingly sophisticated capabilities to attackers.

OpenAI therefore appears to be positioning cybersecurity AI as both a product opportunity and a security challenge.

Sam Altman Teased a “New Thing”

Ahead of DevDay, Sam Altman teased the event on social media, writing:

“Pretty excited for DevDay tomorrow. We have found a new thing.”

The post generated substantial attention ahead of the conference.

With OpenAI expected to introduce multiple products and platform updates, the comment has naturally fuelled speculation about what the company considers its next major product category.

A Broader Developer Platform

DevDay is not only about consumer-facing products.

OpenAI’s official event description emphasises APIs, tools, technical sessions and hands-on development.

That suggests developers should also expect improvements to the infrastructure surrounding OpenAI models.

Potential areas of interest include:

• Agent development
• Tool use
• Computer interaction
• API improvements
• Enterprise AI
• Cybersecurity
• Workflow automation
• Developer tooling

The bigger question is how these pieces fit together.

OpenAI’s Competitive Challenge

OpenAI helped establish the modern general-purpose AI API market.

But the competitive landscape has changed significantly.

Developers increasingly have access to complete AI products rather than simply raw model endpoints.

Google, Microsoft, Meta and specialist agent companies are all pushing towards systems that can perform tasks rather than merely generate responses.

That means OpenAI must demonstrate that its platform can support not only powerful models but also reliable, useful and secure agents.

What Developers Should Watch Today

Several announcements could have major implications for developers.

The most important areas to watch include:

AI agents: Whether OpenAI introduces a new persistent or autonomous agent platform.

GPT-6 Cyber: Whether the cybersecurity model receives an official preview.

API tools: New capabilities for developers building agentic applications.

Computer use: Improvements to OpenAI’s ability to let models interact with software.

Enterprise AI: New tools for organisations deploying AI at scale.

Safety: How OpenAI plans to control increasingly autonomous systems.

Final Thoughts

OpenAI DevDay 2026 arrives at a turning point for the AI industry.

The central question is no longer simply which company can build the most capable general-purpose model.

It is increasingly about which company can turn those models into reliable agents that can actually accomplish useful work.

OpenAI has already demonstrated advanced computer-use capabilities through GPT-6 Astra. Now developers will be watching to see how the company turns that technology into a broader platform.

The reported “O” agent remains unconfirmed, while GPT-6 Cyber is a reported upcoming product rather than a confirmed DevDay launch.

But both rumours point towards the same direction: OpenAI is moving deeper into the agent era.

And today’s DevDay keynote could reveal just how far the company plans to take that shift.