At OpenAI’s DevDay 2026 on September 29, the company unveiled autonomous “Dots” agents, the faster GPT-6.1 Sol model, an Ultrafast speed tier, Private Intelligence, Codex upgrades, new Decisions and Agents APIs, Bedrock Managed Agents, and an expanded ChatGPT platform for builders.
GGLOBAIMODELS DESKSHARE
At OpenAI’s DevDay 2026 on September 29, the company unveiled autonomous “Dots” agents, the faster GPT-6.1 Sol model, an Ultrafast…
Share this post
Short answer: At OpenAI’s DevDay 2026 on September 29, the company unveiled autonomous “Dots” agents, the faster GPT-6.1 Sol model, an Ultrafast speed tier, Private Intelligence, Codex upgrades, new Decisions and Agents APIs, Bedrock Managed Agents, and an expanded ChatGPT platform for builders.
DevDay 2026 announcements overview
OpenAI’s yearly developer gathering took place on September 29, 2026, and the stage was packed with more than twenty fresh announcements that touched everything from flagship products to the underlying models and brand-new ways for people and AI to work together. The vibe was clear: these releases are meant to carve out more time for meaningful work and to let us chase ideas that once felt out of reach.
OpenAI autonomous agents for ongoing chores
A big part of the talk centered on autonomous agents that can shoulder ongoing chores for us. They’re called Dots-always-on helpers that learn what matters to each person, operate in the background and free up mental bandwidth for higher-level thinking. Right now Dots are rolling out to Pro and Business Premium subscribers in eligible markets, while Enterprise, Education and Healthcare customers can grab a beta once their workspace admin flips the switch.
Model upgrades also stole the spotlight. GPT-6.1 Sol debuted as the next step up from the GPT-6 Sol family, delivering sharper performance on agentic coding, computer use and professional workloads while charging only a fifth of the usual token price for input and output. Alongside that, a brand-new Ultrafast speed tier promises up to eight-times faster token generation in Codex and up to six-times faster in the API-perfect for any scenario where latency can’t afford to linger.
Private Intelligence security and privacy features OpenAI
Security and privacy got their own dedicated segment with the Private Intelligence offering. This service lets organizations run frontier AI models with a guarantee that their data isn’t retained, thanks to zero-data-retention processing for safety reviews. A preview of Private Inference, slated for later this year, marries confidential computing with verifiable controls to add another layer of protection for sensitive information during model inference.
Codex ecosystem updates new APIs ChatGPT shared surface
Developers got a heap of updates to the Codex ecosystem. Codex can now run locally, on a remote phone or in the cloud from any device, complete with reusable environments that streamline setup and enforce shared permissions. The Codex CLI picked up voice-driven task initiation, a fresh /agents view for delegating work and an improved terminal layout that’s easier on the eyes during long sessions. Inside the ChatGPT desktop app there’s a code review feature that lets users summarize changes, inspect diffs and consult Codex before pushing feedback to GitHub or GitLab. On top of that, Codex Security Cloud continuously scans GitHub repositories, removes duplicates and readies fixes-all without needing a local machine to stay awake.
New APIs opened doors for more agentic applications. The Decisions API narrows model focus to a predefined set of questions with limited answers, enabling classification, routing or action selection based on text or image input. The Agents API now supports direct computer use, so builders can craft agents that interact with software, search tools, call functions and compress context while OpenAI handles the underlying infrastructure. A collaboration with Amazon birthed Bedrock Managed Agents, which package the Agents API capabilities for native deployment inside AWS environments.
Finally, OpenAI flung ChatGPT open as a shared surface where humans and agents can coexist, and where developers can launch native experiences that reach the platform’s 1.2 billion weekly users. Plugin extensions let creators drop in interactive panels, sidebars and custom viewers for various file types, while tweaks to plugin creation, submission and discovery aim to lower the barrier to entry and surface the right tools more effectively.
For anyone building with-or simply using-AI, these releases translate into more flexibility to automate routine tasks, higher performance for latency-sensitive work and stronger guarantees around data protection. To make the most of them, developers should play with the new agent frameworks, give the Ultrafast tier a spin for speed-critical pipelines, explore Private Intelligence for compliance-heavy projects, refresh their Codex workflows with the cloud and CLI upgrades, and consider building or updating plugins to tap into the expanded ChatGPT surface. Keeping an eye on the official docs and community forums will help ensure the latest capabilities slide smoothly into existing projects.
Frequently asked questions
When was DevDay 2026 held?
OpenAI’s yearly developer gathering took place on September 29, 2026, and the stage was packed with more than twenty fresh announcements that touched everything from flagship products to the underlying models and brand-new ways for people and AI to work together.
What are Dots and who can access them?
Dots are autonomous, always-on helpers that learn each user’s priorities, run in the background, and free up mental bandwidth for higher-level thinking. They are currently available to Pro and Business Premium subscribers in eligible markets, while Enterprise, Education, and Healthcare customers can access a beta once their workspace admin enables it.
What is GPT-6.1 Sol and how is it priced?
GPT-6.1 Sol debuted as the next step up from the GPT-6 Sol family, delivering sharper performance on agentic coding, computer use, and professional workloads while charging only a fifth of the usual token price for input and output.
What is the Ultrafast speed tier and its benefits?
A brand-new Ultrafast speed tier promises up to eight-times faster token generation in Codex and up to six-times faster in the API, ideal for latency-sensitive scenarios.
OpenAI unveiled Dots, an always-on AI assistant, at its DevDay keynote on September 29, 2026. Powered by GPT-6 Astra, each Dot runs on its own cloud computer with a web browser and access to over 4,000 apps, positioning it as a direct rival to Meta’s Muse.
AMD has agreed to acquire AI startup World Labs for $8.2 billion, with the deal expected to close before the end of 2026 pending regulatory approval. The purchase aims to strengthen AMD’s artificial-intelligence portfolio and narrow the gap with Nvidia by adding World Labs’ world-modeling technology, synthetic-data tools, and the expertise of founders Fei-Fei Li, Justin Johnson, Christoph Lassner, and Ben Mildenhall.
AMD has agreed to acquire artificial-intelligence startup World Labs for $8.2 billion, with the deal expected to close by year-end pending regulatory review. The purchase aims to strengthen AMD’s position in the AI hardware market against Nvidia’s lead.
NO COMMENTS YET
Comments are open. Have a thought or a question? Share it below.