← Back to all tools
TOOL · CODING / AGENTS

Codex

Codex is the OpenAI coding agent we list separately from ChatGPT because the experience is a separate app and developer workflow. It can read a repo, edit files, run commands and tests, and move work toward a reviewed branch instead of staying as advice in a chat.

Our rating
Pricing
ChatGPT plan
Last re-tested
May 1, 2026
VERDICT
Show it separately from ChatGPT. The billing may be bundled, but the job is different: Codex is where repo work, tests, and implementation loops happen.

Best for

  • Repo work where the agent needs to inspect files, edit code, and run the test loop
  • Turning scoped specs into implementation branches, bug fixes, migrations, or cleanup passes
  • Parallel agent work across projects when we need multiple code tasks moving at once
  • Keeping code execution separate from general ChatGPT prompts and content work

Skip if

  • The person using it does not want to review diffs, terminal output, or code changes
  • The task touches production access, destructive git work, or credentials without explicit guardrails
  • You only need a simple coding explanation; ChatGPT or Claude may be faster

Pricing breakdown

Plan
Cost
What you get
Included access
ChatGPT plan
OpenAI lists Codex as included with Plus, Pro, Business, Enterprise/Edu plans; usage still depends on plan limits.
Free / Go access
Limited time
OpenAI currently describes Free and Go access as a limited-time inclusion, so do not anchor long-term workflow decisions on that tier.
Business use
Varies by plan
For heavier coding-agent work, evaluate plan limits, workspace controls, GitHub permissions, and cloud/local execution needs.

Playbooks that use Codex