OpenAI DevDay 2026: First announcements bring GPT-6.1 Sol, Codex Cloud and native plugins
OpenAI has published its [DevDay 2026 recap](https://openai.com/index/devday-2026-recap/) while the event is still in progress. It lists more than 20 announcements spanning models, Codex, APIs, plugins and team workflows. This report covers the first official announcements; a separate post-event recap will check the full program and any later technical details.
GPT-6.1 Sol: a lower-cost model for agentic work
The headline model is **GPT-6.1 Sol**, which OpenAI positions for agentic coding, computer use and professional tasks. Standard API pricing is **$2 per million input tokens, $0.10 per million cached input tokens and $10 per million output tokens**. OpenAI says standard input and output rates are one fifth of GPT-6 Astra's. Its published DeepSWE and OSWorld results suggest major gains over GPT-6 Sol, but these are [OpenAI's own evaluations](https://openai.com/index/introducing-gpt-6-1-sol/), not independent production results.
The API model ID is `gpt-6.1-sol`. OpenAI says the model is available now through the API and in ChatGPT Work and Codex for Plus, Pro, Business, Enterprise and Edu users. **It is not yet available in Chat.** That distinction matters if a team is planning a rollout across API applications and the consumer chat interface. [Source: GPT-6.1 Sol announcement](https://openai.com/index/introducing-gpt-6-1-sol/)
OpenAI also introduced **Ultrafast**, a premium speed tier separate from its existing Fast mode. It claims up to eight times faster token generation in Codex, up to 300 tokens per second, and up to six times faster in the API. **GPT-6 Astra Ultrafast is available now** in the API and to Pro 500 and Enterprise users in Work and Codex. **GPT-6.1 Sol Ultrafast is coming soon.** The [API guide](https://developers.openai.com/api/docs/guides/ultrafast-mode) recommends persistent WebSocket connections for workflows with repeated tool calls and warns that the speed carries a higher cost.
Codex moves from a local session to a reusable cloud workspace
**Codex Cloud** can run coding tasks in a prepared environment while the developer's computer is asleep. Teams can reuse a setup containing repositories, dependencies, tools and approved access, while each task keeps its own workspace. Results can be reviewed from desktop, web or mobile. The [Codex Cloud guide](https://learn.chatgpt.com/docs/cloud) describes the setup and review flow; OpenAI lists access on Plus, Pro, Business, Healthcare, Education and Enterprise plans.
The Codex CLI gains voice steering, an `/agents` view, improved session resumption and worktree handling. The desktop app gains a code review view with summaries and diffs, and automatic first-pass reviews can run in the cloud. These changes make remote execution and review a larger part of the coding workflow, rather than a one-off remote task. [Source: DevDay recap](https://openai.com/index/devday-2026-recap/)
**Codex Security Cloud** adds repository-wide GitHub scans on demand or on a schedule, checks new commits, deduplicates findings and prepares fixes remotely. OpenAI says it includes access to Daybreak Blue models and is available to Pro, Business, Enterprise and Edu users on desktop and web. This is a security workflow across a repository, distinct from routine code review. [Source: DevDay recap](https://openai.com/index/devday-2026-recap/)
ChatGPT plugins gain their own product surfaces
OpenAI is opening more of ChatGPT's interface to developers. A plugin can now have an entry in the sidebar, an interactive panel beside a conversation, product settings, and viewers or editors for supported files. The [extension documentation](https://developers.openai.com/plugins/build/extensions) also describes deep links, richer forms and context sharing with the model. A plugin is the installable package: it can contain skills, an MCP server or both. The MCP server supplies live tools and data; skills guide repeatable workflows. [Source: plugin architecture](https://developers.openai.com/plugins/concepts/plugins)
OpenAI is improving Plugin Creator, submission feedback and discovery. It also says supported plugins can run inside **Sites** for a team, with each person's connected data and permissions. **MCP Events** gives plugins a way to trigger ChatGPT automations when a connected service emits an event, such as a new message or document comment. The [MCP Events guide](https://developers.openai.com/plugins/build/mcp-events) describes webhook delivery against a draft protocol; deployment status and compatibility need checking before a production integration. The recap lists plugin extensions across plans, while detailed docs say web support for Free and Go is still coming and composer mentions require the desktop app. [Sources: DevDay recap](https://openai.com/index/devday-2026-recap/), [plugin extensions](https://developers.openai.com/plugins/build/extensions)
Agents API and business workflows
The **Agents API** now supports computer use, allowing agents to interact with software. OpenAI also lists multi-agent orchestration, tool search, tool calling and context compaction. The API is in public beta, with usage charged through the underlying models and tools rather than a separate Agents API fee. OpenAI and Amazon announced Bedrock Managed Agents for teams building these agents within AWS. [Sources: Agents API announcement](https://openai.com/index/introducing-the-agents-api/), [DevDay recap](https://openai.com/index/devday-2026-recap/)
A separate **Decisions API** uses Luna to choose from a finite set of developer-defined answers using text or image context. OpenAI suggests classification, routing and next-action selection. It is in **limited preview today**, with wider release planned in the coming days. For sensitive workloads, OpenAI also detailed Zero Data Retention with Private Safety Processing; **Private Inference is a preview planned for this fall**. [Sources: DevDay recap](https://openai.com/index/devday-2026-recap/), [Private Safety Processing guide](https://developers.openai.com/api/docs/guides/private-safety-processing)
What to watch after the closing session
The wider announcements include always-on dots, ChatGPT Space and Pages for collaborative work, shared team tasks, @ChatGPT in Slack and Teams, Sign in with ChatGPT, a Pro 500 tier and an enterprise Marketplace. Collaborative slides are promised for the coming weeks. These launches indicate where OpenAI expects agents and plugins to be used, but the plan restrictions and stages of rollout differ. We will publish a separate, fully checked DevDay wrap-up after the event closes. [Source: DevDay recap](https://openai.com/index/devday-2026-recap/)