Codex CLI Scripting

🚨 CRITICAL: NEVER USE OLDER MODELS 🚨

ONLY use these models:

• gpt-5.3-codex — for coding tasks (code review, debugging, writing code, etc.)
• gpt-5.2 — for everything else (research, writing, analysis, etc.)

NEVER use: o3, o4-mini, gpt-5.1-codex-max, gpt-5.2-codex, or any other older model.

ALWAYS set reasoning effort explicitly: -c model_reasoning_effort="high" (or "xhigh")

This applies to ALL calls regardless of task type. Non-negotiable.

Quick Reference

Headless Command Syntax