Bug: @openai/codex 0.115.0 fails on macOS Apple Silicon with missing optional dependency @openai/codex-darwin-arm64
Open 💬 3 comments Opened Mar 16, 2026 by TholeG
What version of Codex CLI is running?
0.115.0 fails immediately after install.
Confirmed workaround: 0.114.0 installs and runs correctly on the same machine.
What subscription do you have?
Plus
Which model were you using?
_No response_
What platform is your computer?
macOS on Apple Silicon (darwin arm64)
What terminal emulator and version are you using (if applicable)?
zsh on macOS
What issue are you seeing?
Installing the current npm release succeeds, but the CLI immediately fails at runtime because the platform-specific optional dependency is missing.
npm install -g @openai/codex
# changed 2 packages in 7s
codex --version
Then:
file:///Users/<user>/.npm-global/lib/node_modules/@openai/codex/bin/codex.js:100
throw new Error(
^
Error: Missing optional dependency @openai/codex-darwin-arm64. Reinstall Codex: npm install -g @openai/codex@latest
What steps can reproduce the bug?
node -v
# v22.15.0
npm -v
# 11.11.0
uname -m
# arm64
node -p 'process.platform + " " + process.arch'
# darwin arm64
npm uninstall -g @openai/codex
npm install -g @openai/codex
codex --version
What is the expected behavior?
codex --version should print the installed CLI version.
Additional information
The npm debug log for the failed install contains:
verbose reify failed optional dependency /Users/<user>/.npm-global/lib/node_modules/@openai/codex/node_modules/@openai/codex-darwin-arm64
A few extra observations from local debugging:
@openai/codex@0.115.0installs the wrapper package, but@openai/codex-darwin-arm64is not present in a usable form afterwards.@openai/codex@0.114.0installs correctly on the same machine andcodex --versionworks.- The published variant version appears to exist on npm (
0.115.0-darwin-arm64), so this looks like a packaging / optional dependency regression in0.115.0on macOS ARM64.
Current workaround:
npm install -g @openai/codex@0.114.0This issue has 3 comments on GitHub. Read the full discussion on GitHub ↗