OpenAI Codex CLI v0.142.0 Changelog
What changed in Codex CLI v0.142.0, released 2026-06-22 β explained in plain English.
- Smarter Plugins & Features: Access categorized plugins with automatic installation suggestions, safe web search restricted to pre-approved sites, and time-awareness features.
- Performance & Bug Fixes: Experience faster startup times, cleaner logs, and fixes for Linux screen freezes, connection drops, and cross-platform file path issues.
Want a heads-up when OpenAI Codex CLI updates?
We email you the moment a new OpenAI Codex CLI version drops β plus plain-English release notes like these. No spam.
This site is protected by reCAPTCHA and the Google Privacy Policy and Terms of Service apply.
This release is all about making the Codex command-line tool smarter, faster, and much more reliable. It introduces better ways to manage your usage limits, organizes your plugins (add-ons that give the AI extra tools) more neatly, and helps different AI agents work together without crashing. If you use Codex to help you write code, this update means fewer interruptions, quicker startup times, and more control over how the AI behaves.
What's new
- Usage limit resets: You can now use the
/usagecommand to see and redeem credits that reset your usage limits, complete with confirmation screens so you don't use them by accident. - Smarter plugin organization: The
/pluginscommand now groups tools into clear categories (like OpenAI Curated, Workspace, and Shared with me). Codex can also suggest and install helpful plugins automatically during your chat. - Token budgets: You can now set a limit on "tokens" (the basic units of text and code the AI reads and writes). This keeps the AI from running wild, warns you when you are running low, and stops the AI if it runs out of budget.
- Control over AI teamwork: If you use multiple AI agents to solve a problem, you can now decide exactly how they hand off work to each otherβwhether they do it automatically, ask you first, or don't do it at all.
- Safe web search: A new search mode lets Codex search the live web while restricting it to safe, pre-approved websites.
- Time awareness: Codex can now check the current time and receive scheduled time reminders, helping it stay on track with time-sensitive tasks.
Improvements
- Faster startup: Codex now starts up and loads your sessions much faster. It does this by delaying unnecessary network checks, pre-loading the AI model in the background, and loading plugin skills all at once instead of one by one.
- Cleaner logs: The app now saves less unnecessary background data, keeping your log files smaller and cleaner.
Fixes
- Linux screen freeze: On Linux, if you paused the app using
Ctrl+Zand brought it back using thefgcommand, the visual interface would sometimes break. This has been fixed. - Better connection recovery: If your internet drops for a second, Codex's background servers will now automatically reconnect and retry sending your work instead of crashing.
- Cross-platform fixes: Fixed issues where remote coding environments would get confused by different operating systems (like Windows vs. Mac/Linux) when handling file paths and terminal shells.
- Smoother plugin installs: Fixed several bugs that caused plugins to fail to install, and made error messages much clearer when a download goes wrong.
- Honest AI error reporting: If a helper AI agent fails at a task, it now properly reports the error to the main AI agent instead of pretending the job was completed successfully.
- Saved chat threads: Fixed a bug where certain "goal-first" chat threads were not being saved or showing up in your search history.
Breaking changes
No breaking changes in this release.
How to upgrade
Update to OpenAI Codex CLI v0.142.0 by running:
codex --upgrade
This is our plain-language summary. Read the complete, official notes on the OpenAI Codex CLI official changelog β.
