Summary
OpenAI announced more than 20 updates at DevDay 2026, including an Agents API that can interact with software, new Codex workflows and collaborative features in ChatGPT. The announcements also include new model and speed options, with access varying by plan and feature.
OpenAI announced more than 20 updates at DevDay on September 29, 2026, spanning ChatGPT, Codex, its models and tools for developers. A central change is that developers can now build agents that interact with software through the Agents API, while new ChatGPT features bring people, agents and shared work into the same workspace.
OpenAI also said developers can launch native experiences within ChatGPT, which it describes as a shared surface for people and agents, to its collective 1.2 billion weekly users.
Contents
Agents and developer workflows
The Agents API now supports computer use, allowing developers to build agents that interact with software to complete tasks. OpenAI says the API also brings in Codex’s multi-agent capabilities, tool search, tool calling and context compaction. OpenAI operates the underlying infrastructure. The feature is available through the API and in Codex and ChatGPT Work on Pro 500 and Enterprise plans.
A separate Decisions API is designed for tasks with a fixed set of possible answers. Developers provide context as text or images, and the API returns answers for uses such as classifying content, routing requests or selecting an agent’s next action. It was announced in limited preview, with a broader release planned in the coming days.
Codex is also gaining more ways to run and manage development work. Developers can run it on a computer, remotely from a phone or in the cloud, using reusable environments with shared settings and permissions. In the Codex CLI, voice controls and a new agents view are intended to help users steer tasks and manage several at once.
Other Codex updates include code review in the ChatGPT desktop app, where users can inspect summaries and diffs and ask Codex about potential issues in GitHub pull requests or GitLab merge requests. Codex Security Cloud can scan GitHub repositories on demand or on a schedule, investigate findings and prepare fixes in the cloud.
ChatGPT as a shared workspace
OpenAI is opening up some of the tools used to build ChatGPT features through plugin extensions. Developers can create interactive panels in the ChatGPT sidebar and viewers for supported file types. Users choose which plugins to use and approve the access each receives.
ChatGPT Space is a shared area where teammates, ChatGPT and an OpenAI agent called a dot can work with shared knowledge. Pages add collaborative documents where users can write, research, create charts or images, and invite teammates to contribute. OpenAI also announced recurring team tasks that can use connected tools and run on a schedule or in response to events, such as a new email or Slack message.
Several features have narrower access or staged availability. ChatGPT Space is available to Pro, Business and Enterprise users on desktop and web, while mobile creation and editing are still to come. Collaborative slides were announced for release in the coming weeks. Dots, which OpenAI describes as always-on agents for ongoing responsibilities, are available on Pro and Business Premium in eligible markets; Enterprise, Education and Healthcare users can try a beta only if their workspace administrator enables it.
Models, speed and business controls
OpenAI introduced GPT-6.1 Sol, describing it as a major upgrade to GPT-6 Sol for agentic coding, computer use and professional work. The company says it offers near-Astra intelligence at one-fifth of its standard input and output token prices. It is available to API, Plus, Pro, Business, Enterprise and Education users.
The new Ultrafast tier is aimed at workloads where speed is a priority. OpenAI says it provides up to 8× faster token generation, at 300 tokens per second, in Codex and up to 6× in the API. GPT-6 Astra Ultrafast is available in the API and in ChatGPT Work and Codex on Pro 500 and Enterprise plans; GPT-6.1 Sol Ultrafast is planned for a later release.
For businesses, OpenAI also announced Private Intelligence features. Zero Data Retention with Private Safety Processing is designed to automate safety reviews without giving OpenAI personnel access to the underlying content. A preview of Private Inference, which OpenAI says combines confidential computing with verifiable controls, was planned for later in the fall.