OpenAI unveils 20 updates with new agents & GPT-6.1

OpenAI unveiled more than 20 product updates across ChatGPT, its models and developer tools at its DevDay event in San Francisco. The announcements included new AI agents called Dots, a shared workspace called ChatGPT Space and a new model named GPT-6.1 Sol.

Dots are always-on agents that run on a cloud computer and work across a user's existing context and connected apps. They are designed for ongoing tasks and can make progress in the background over time while learning from feedback.

ChatGPT Space is a shared environment for teams using ChatGPT. It brings together teammates, ChatGPT and a Dot in one place, while organising chats, uploaded files and shared work so users can continue from the same context.

New model

Alongside those tools, OpenAI launched GPT-6.1 Sol, describing it as an upgrade to GPT-6 Sol. The model is aimed at coding, computer use and professional work, and offers what OpenAI called near-Astra intelligence at one-fifth of Astra's price.

The pricing claim suggests OpenAI is trying to widen access to more advanced models by lowering the cost of frequent use. That matters as developers and businesses weigh model performance against the expense of deploying AI systems across daily workflows.

Another update lets ChatGPT subscribers use their existing accounts in third-party products. Through a new sign-in option, Plus and Pro users can access OpenAI models in partner tools without setting up an API key.

The feature is launching with 16 partners, including Cognition, Notion, Lovable, Vercel and OpenClaw. The move could help OpenAI extend the reach of its subscription products beyond its own apps and website.

Developer push

OpenAI also expanded the ways developers can build software inside ChatGPT and Codex. Plugin extensions can now take a place in the sidebar, show an interactive panel next to a conversation or create a viewer for a file type.

OpenAI also said it has improved the process for creating, submitting and discovering plugins. That suggests it is trying to make ChatGPT not only a destination for end users but also a platform where outside developers can distribute software.

The breadth of the updates shows OpenAI continuing to push its products beyond chat interfaces and into broader workplace and software tasks. The additions span individual subscriptions, team collaboration, model economics and developer distribution, all central areas in the race to turn generative AI tools into everyday business products.

Its description of Dots also points to a stronger focus on autonomous systems that can take on work with less direct prompting. By tying those agents to existing apps and context, OpenAI is moving further towards software that manages tasks over time rather than responding only to one-off requests.

The event drew 2,500 developers, according to OpenAI. The scale of the gathering underlines the company's effort to keep software makers engaged as rivals compete for developers with lower model prices, new coding tools and broader commercial terms.

While OpenAI gave only limited operational detail, the package outlines a strategy built around three layers: more persistent AI agents, cheaper access to stronger models and tighter links between ChatGPT and other software. It also indicates that OpenAI sees collaboration and workflow management as areas where AI products can move deeper into office use.

For developers, the practical significance may lie in how easily these products can be integrated into existing tools and habits. For businesses, the key questions are likely to centre on cost, data control, the reliability of autonomous actions and whether teams are willing to make ChatGPT a shared layer for day-to-day work.

OpenAI said the launch covered more than 20 updates across ChatGPT, models and new ways of working.