Codex CLI 0.166.0 adds keyboard transcript selection
Updated Codex CLI
Summary
Codex CLI 0.166.0 (October 29, 2026) adds keyboard transcript selection and search, codex doctor --network-only, codex queue --resume and resume fixes.
- Tool
- Codex CLI
- Category
- Coding agents
- Date
- Open source
- Yes
- MCP
- Yes
- Pricing
- Not stated in the sources we cited.
OpenAI tagged Codex CLI 0.166.0 on October 29, 2026, two days after 0.165.0. It is a TUI and reliability release: you can now select and search the transcript from the keyboard, and resuming or forking a thread no longer loses approval settings or reasoning-token counts. The same build reached npm as @openai/codex@0.166.0 the same evening (UTC), under the latest tag.
Release facts
| Fact | Detail |
|---|---|
| Product | Codex CLI, OpenAI's local coding agent |
| Version | 0.166.0 (GitHub tag rust-v0.166.0) |
| Release date | October 29, 2026 (GitHub release and npm upload) |
| Previous stable | 0.165.0, October 27, 2026 |
| License | Apache-2.0, per the GitHub API |
| Install | npm install -g @openai/codex@0.166.0, brew install --cask codex, or the install script in the README |
What's new
- Keyboard transcript selection. Word, line and page motions, built-in search and configurable shortcuts for picking text out of the transcript (PRs 57221 to 57224).
codex doctor --network-onlyruns only the configuration, connectivity and WebSocket checks, which is the part you need when a proxy or firewall is the suspect (PR 57606).codex queue --resumeresumes unloaded threads after you queue a message to them (PR 57232).- Safety blocks explain themselves. The TUI now says why a request was blocked and offers safer follow-up prompts you can select (PRs 57363 and 57365).
- GPT-6.1 Terra is in the bundled, Vertex AI and Amazon Bedrock model catalogs (PR 57293). 0.165.0 had already added it to the default catalogs and added a built-in Vertex AI provider, set with
model_provider = "gcp". - Markdown renders the same way in async questions, choices, answers and transcript replies.
What's fixed
- Resume and fork keep your settings. Approval and permission choices you had acknowledged survive a resume or fork (PR 57227), and reasoning-token accounting carries over, so a resumed thread no longer compacts early (PR 57839).
- Unarchiving a parent thread restores the descendant threads it spawned.
- Killed composer text comes back with its local image attachments.
- Reconnects: the TUI recovers more cleanly, and
/quitand/exitwork while offline. - macOS standalone installs clean up correctly when files are in use or paths contain non-ASCII characters.
- App-server startup and shutdown races are fixed, and finished runtimes are released promptly.
Questions
Is anything breaking? The release notes list no breaking change. Under the hood, development and release builds move to Node.js 24.22.1 and Rust 1.95.0, and the standalone executor now ships inside the CLI and app-server packages. That matters only if you build Codex from source.
Should I take 0.166.0 or wait?
Take it if you resume or fork threads often, since the two resume fixes address settings and context loss. The 0.166.0-alpha builds published on npm between October 27 and 29 were previews of this release; skip them now that the stable build is out.
Where does it fit next to other terminal agents? See Claude Code vs Codex for the closest comparison, and the coding agents category for OpenCode and Gemini CLI.
Alternatives
| Tool | License | Pricing note |
|---|---|---|
Codex CLIThis launchCodex CLI is OpenAI's Apache-2.0 coding agent that runs locally in your terminal, signed in with a ChatGPT plan or an OpenAI API key. | Apache-2.0 | The CLI is Apache-2.0. Usage comes from a ChatGPT plan or is billed at API rates when you sign in with an API key. |
Claude CodeClaude Code is an agentic coding tool that runs in your terminal and works on your codebase. | — | Not stated |
OpenCodeOpen-source coding agent from the anomalyco/opencode repository, MIT-licensed, with a terminal client and a site at opencode.ai. | MIT | The client is MIT-licensed. Model access is separate; this page does not quote a hosted plan. |
Gemini CLIGoogle's Apache-2.0 terminal coding agent for Gemini models. Version 0.67.0 (November 4, 2026) adds a models command and sandbox IDE diffs. | Apache-2.0 | The CLI is free under Apache-2.0. Model use is set by how you sign in: a free Google account tier, a Gemini API key (free or paid), Vertex AI, or a Google AI Pro, Ultra or Code Assist plan. |
Questions
What is Codex CLI?
The openai/codex README describes Codex CLI as a coding agent from OpenAI that runs locally on your computer. The same repository points to an IDE extension, the desktop app (codex app) and the cloud agent Codex Web at chatgpt.com/codex.
Is Codex CLI open source?
Yes. The GitHub API reports Apache-2.0 for openai/codex, and the README states the repository is licensed under Apache-2.0. Most of the code is Rust.
How do I install Codex CLI?
The README lists an install script for macOS and Linux, a PowerShell script for Windows, npm install -g @openai/codex and brew install --cask codex. Then run codex and choose Sign in with ChatGPT, or set it up with an API key.
Do I need a ChatGPT subscription to use Codex CLI?
No. The README recommends signing in with a ChatGPT plan, but the pricing page lists an API key option that runs Codex in the CLI, SDK or IDE extension, billed at API rates and without cloud features such as GitHub code review or Slack.
Filed on Agent Radar. Claims in this note are limited to the sources above.