Source · release
Codex CLI 0.165.0
Summary
Codex CLI 0.165.0 was released on 2026-10-27. It adds a built-in Google Cloud Vertex AI provider, turns Memory v2 on by default for new profiles (with import of memories from other agents), and lets you change reasoning effort mid-conversation. It also adds GPT-6.1 Terra to the model catalogue and better session navigation across worktrees.
Key points
- Vertex AI provider: set
model_provider = "gcp"in the Codex config to run through Google Cloud Vertex AI without a custom provider block. - Memory v2 by default: fresh profiles get Memory v2 switched on. It can import memories from other coding agents, so you can carry context over when switching tools. Existing profiles keep their current setting.
- Effort mid-conversation: on supported models, reasoning effort can now be changed during a conversation by default.
- GPT-6.1 Terra is in the model catalogues.
- Worktrees: the session picker gains a repository filter, and the agent command centre groups sessions by directory, which helps when you work across linked worktrees.
- App-server API: experimental attached thread forks.
- Update with
npm install -g @openai/codex@0.165.0(or your package manager).
Why it matters
Memory import lowers the cost of trying Codex next to Claude Code. The Vertex provider makes Codex usable for teams whose model access goes through Google Cloud.