OpenAI gives Codex reusable cloud environments that work across devices

OpenAI unveiled new capabilities for its software engineering agent Codex at its DevDay 2026 conference on September 29. The headline addition is reusable cloud development environments that let a developer work with Codex on a computer, from a phone, or in the cloud.

Instead of treating every cloud task as an isolated remote sandbox, OpenAI is making those environments persistent and configurable. Teams get a shared workspace with approved settings and permissions, which helps tasks start faster. Codex in the cloud is available on Plus, Pro, Business, Enterprise, Edu, and Healthcare plans.

A revamped CLI and code review in the desktop app

The updated Codex CLI can start and steer tasks by voice. A new /agents view makes delegating work easier while showing several tasks at once, and OpenAI improved prompt editing, session resuming, and Git worktrees, with a cleaner terminal interface. The CLI is available on all plans.

Inside the ChatGPT desktop app, Codex powers a new code review view: developers can read summaries, go through diffs, and ask Codex about possible problems before leaving feedback on GitHub pull requests or GitLab merge requests. Automatic reviews let Codex do a first pass while the developer is away.

Codex Security Cloud scans repositories

Codex Security Cloud targets security teams. It scans GitHub repositories on demand or on a schedule and keeps checking new commits, then investigates findings, filters out duplicates, and prepares fixes in the cloud even when the laptop is closed. Cloud users also get the Daybreak Blue models without applying separately. It is available for Pro, Business, Enterprise, and Edu.

Agents API, Decisions API, and Ultrafast

OpenAI is adding Computer Use to its Agents API, letting agents operate software on their own. The API also includes Codex's multi-agent features, Tool Search, and Context Compaction, which automatically condenses long contexts. Through Amazon's Bedrock Managed Agents, developers can build OpenAI agents that run entirely on AWS.

The new Decisions API is built for real-time decision-making. It uses GPT-6 Luna to answer user-defined questions with a predefined set of answers, taking context as text or images, to classify content, route requests, or pick an agent's next step. OpenAI says it returns a decision in 150 milliseconds, about ten times faster than GPT-6 Luna through the regular API. It launches as a limited preview.

Ultrafast, a premium tier, speeds up token generation in Codex by up to eight times, to 300 tokens per second, and up to six times in the API at six times the standard rate. For GPT-6 Astra that means $60 per million input tokens and $300 per million output tokens. It is available now in the API and for Enterprise customers and subscribers to the new $500-a-month Pro 500 plan.

Sources: TechCrunch · The Decoder