OpenAI held its 2026 DevDay conference at Fort Mason in San Francisco on September 29, bringing developers and technical founders together for announcements spanning AI models, developer tools and agent capabilities. The event focused on new ways for developers to build, run and collaborate on AI-powered applications.
Key Takeaways
- OpenAI held DevDay 2026 at Fort Mason in San Francisco on September 29.
- The event featured announcements involving AI models, developer tools, Codex and AI agents.
- OpenAI described the 2026 conference as its biggest DevDay to date.
- Developers, technical founders, researchers and technical leaders attended sessions, demonstrations and workshops.
- New releases expanded OpenAI’s tools for coding, agent development, collaboration and application workflows.
OpenAI DevDay Opens in San Francisco
OpenAI brought its annual developer conference to Fort Mason in San Francisco on September 29, with programming centered on developers and technical builders working with artificial intelligence. The schedule included an opening keynote, technical sessions, hands-on demonstrations, workshops and opportunities to interact with OpenAI teams.
OpenAI described the 2026 conference as its biggest DevDay to date and said the event included more than 20 major announcements covering ChatGPT, Codex, models and new ways of working with AI.
The in-person audience included engineers, technical founders, researchers and technical leaders actively building with AI or preparing to develop AI applications. OpenAI also livestreamed the opening keynote and said additional sessions would be recorded and published after the event.
San Francisco hosted the main DevDay gathering, while OpenAI also announced DevDay Exchanges in Bengaluru, Tokyo, Seoul, Paris, Berlin, London, São Paulo and Mexico City later in the year.
The event followed OpenAI’s Mountain View expansion earlier in 2026, which added another major California location to the company’s Bay Area operations.
New AI Models and Developer Tools Announced
OpenAI introduced GPT-6.1 Sol at DevDay as the latest model in the GPT-6 series, with improvements for coding, computer use and professional work. The company released the model in the API and began rolling it out across ChatGPT Work and Codex.
The company also announced Ultrafast, a premium speed tier for workloads where faster token generation is important. OpenAI said the tier could provide up to eight times faster token generation in Codex and up to six times faster generation in the API.
For Codex users, OpenAI introduced cloud-based workflows that allow tasks to continue remotely and be accessed across devices. Reusable cloud environments also give teams a way to begin tasks with shared configurations and approved settings.
The Codex command-line interface received additional updates, including two-way voice controls and an agents view for delegating work and tracking multiple tasks.
OpenAI also expanded Codex code review, giving developers tools to examine summaries, diffs and findings and to ask questions about changes before completing a review.
Codex Security Cloud was another DevDay announcement. The service can scan GitHub repositories, investigate potential vulnerabilities and prepare proposed fixes, with scans available on demand or on a schedule.
The announcements arrived as organizations across the state continue integrating artificial intelligence into workplace and research workflows. A separate California AI adoption study examined how companies, universities and other organizations are using AI across professional settings.
AI Agent Capabilities Expand at DevDay
Agents were a central part of OpenAI’s announcements. The company introduced Dots, always-on agents designed to take on ongoing responsibilities, use connected apps and continue work between conversations. Dots are rolling out to eligible Pro and Business Premium users, while Enterprise customers can access a beta when workspace administrators enable it.
OpenAI also expanded the Agents API with computer-use capabilities. The API allows developers to create agents that can interact with software through a user interface while running in OpenAI-hosted environments.
The Agents API also includes hosted execution, memory, tools and multi-agent support, giving developers additional building blocks for applications in which agents can complete multi-step tasks.
The company introduced the Decisions API in limited preview. It is designed for questions with predefined possible answers and can classify inputs, route requests or choose an agent’s next action.
OpenAI also announced Amazon Bedrock Managed Agents, which allows OpenAI-powered agents to run within Amazon Web Services and integrate with AWS resources.
Developers and Technical Founders Attend San Francisco Event
OpenAI designed the San Francisco gathering around hands-on technical participation. The official schedule included sessions on APIs and developer tools, demonstrations, workshops and opportunities for attendees to exchange ideas with OpenAI teams and other builders.
The day began with breakfast before the opening keynote featuring OpenAI CEO Sam Altman. Breakout sessions and programming followed, with a closing session and evening reception completing the schedule.
The San Francisco event also served as the starting point for DevDay Exchanges planned in cities outside the United States, extending the developer-focused program to additional technology communities later in 2026.
OpenAI Details Expanded Tools for AI Application Development
The DevDay announcements extended beyond models and agents into collaboration and application-development tools. OpenAI introduced ChatGPT Space as a shared environment where teams can organize files, Pages and project context while working with ChatGPT and agents.
OpenAI also introduced Pages, editable documents that people and agents can work on together. Pages support writing, research, charts and interactive content, while collaborators can contribute edits and comments within the same workspace.
Collaborative slides were announced for a future release, with OpenAI saying the feature would allow teammates and agents to edit presentations together and leave comments.
The company also announced team-sharing tools and Team Tasks for recurring work. These features are designed to let teams share pages, slides, plugins and spreadsheets while assigning work that can run on a schedule or respond to connected events.
OpenAI further announced ChatGPT integrations for Slack and Microsoft Teams, allowing eligible workplace users to interact with ChatGPT in team conversations and use connected tools subject to administrator and user permissions.
The developer announcements also included plugin support for Sites and MCP Events for plugin automations, allowing connected applications to trigger workflows when specified events occur.
Frequently Asked Questions
What is OpenAI DevDay 2026?
OpenAI DevDay 2026 is the company’s annual developer conference. The 2026 event brought developers and technical builders together in San Francisco for technical sessions, demonstrations, workshops and product announcements.
Where was OpenAI DevDay 2026 held?
The main 2026 DevDay event was held at Fort Mason in San Francisco on September 29, 2026. OpenAI also livestreamed the opening keynote for remote viewers.
What did OpenAI announce at DevDay 2026?
OpenAI announced more than 20 developments involving ChatGPT, Codex, models and agent tools. The announcements included GPT-6.1 Sol, Dots, expanded Agents API capabilities, Codex cloud tools and the Decisions API.
Who attended OpenAI DevDay in San Francisco?
The event was designed for developers and hands-on technical builders, including engineers, technical founders, researchers and technical leaders working with AI or preparing to build AI applications.
What new developer tools did OpenAI introduce?
New tools and updates included Codex cloud workflows, an updated Codex CLI, expanded code review, Codex Security Cloud, the Decisions API and computer-use capabilities in the Agents API.
