AgentComparison
OpenAI · product · coding agent (CLI, IDE, web and desktop app)

OpenAI Codex

OpenAI's coding agent for terminal, editor, cloud and desktop, bundled with ChatGPT plans.

Group: Coding agents

Codex is OpenAI's coding agent. OpenAI's docs describe a CLI that inspects code, edits files and runs commands in your repository, plus an IDE extension, cloud tasks and the ChatGPT desktop app. It is included in ChatGPT plans from Free to Pro and Business, or can be used with an API key at API prices.

Documentation researched, not testedChecked 6 October 2026Vendor documentation, not a test

The decision in plain English

Codex is sold through ChatGPT plans, and the pricing page says ChatGPT Work and Codex share usage, credits and limits. OpenAI lists Free at $0, Go at $8 a month, Plus at $20, Pro from $100 (plans at $100, $200 or $500), Business at $20 per user a month billed annually ($25 monthly) and Enterprise on request. The Free and Go entries list only GPT-6 Luna in the desktop app, subject to rollout, and the page does not state task or message limits, so check what your plan allows before relying on it. The pricing page also says GPT-5.5 retires from Codex on 14 October 2026.

Best fit

Developers who already use or are willing to buy a ChatGPT plan and want one coding agent across terminal, editor and cloud, with sandbox and approval settings they control.

Poor fit

Anyone who needs a fixed, published task allowance on the free tiers, or who needs cloud features such as GitHub code review and Slack on an API-key setup, which the pricing page says are not included.

Documented capabilities

Vendor statements, not verified task outcomes.

  • Terminal agent that inspects code, edits files and runs commands in your repository Source · checked 2026-10-06
  • Open-source Codex CLI under the Apache-2.0 licence Source · checked 2026-10-06
  • Cloud integrations such as automatic code review and Slack on Plus and above Source · checked 2026-10-06
  • OS-enforced sandbox and approval policy, with network off by default Source · checked 2026-10-06

Permissions and human control

  • OpenAI's docs say the agent runs with network access off by default, that locally it uses an OS-enforced sandbox limited to the workspace, and that an approval policy controls when it must stop and ask before acting. Admin controls, SSO and audit logs are listed for Business and Enterprise. These are vendor statements; we did not test them.

Documented controls do not prove that an agent always follows them.

Limits and open questions

  • Plans share usage with ChatGPT Work; the pricing page does not state a fixed number of tasks per plan
  • The Free and Go entries list only GPT-6 Luna at standard speed in the desktop app, subject to rollout
  • API-key use has no cloud features such as GitHub code review or Slack, and is billed at API prices that are separate from subscriptions
  • Pro has three levels ($100, $200, $500) and the pricing page gives access differences only for the top level
  • Model names and availability change; OpenAI says GPT-5.5 retires from Codex on 14 October 2026
  • Described by the vendor; not tested by AgentComparison

Privacy and security

The pricing page says Business data is not used for training by default and links to OpenAI's business-data page. We did not review the consumer-plan data terms for this entry, so check OpenAI's current privacy terms for the plan you use.

Common questions about OpenAI Codex

What is OpenAI Codex for?

Developers who already use or are willing to buy a ChatGPT plan and want one coding agent across terminal, editor and cloud, with sandbox and approval settings they control. The vendor describes it as: Codex is OpenAI's coding agent. OpenAI's docs describe a CLI that inspects code, edits files and runs commands in your repository, plus an IDE extension, cloud tasks and the ChatGPT desktop app. It is included in ChatGPT plans from Free to Pro and Business, or can be used with an API key at API prices.

Who should look elsewhere?

Anyone who needs a fixed, published task allowance on the free tiers, or who needs cloud features such as GitHub code review and Slack on an API-key setup, which the pricing page says are not included.

How much does OpenAI Codex cost?

OpenAI's Codex pricing page lists Free at $0, Go at $8 a month, Plus at $20, Pro from $100 (plans at $100, $200 or $500), Business at $20 per user a month billed annually or $25 monthly with two or more users, Enterprise and Edu on request, and an API-key option billed at API prices. Usage is shared with ChatGPT Work.

What can it access and what needs my approval?

OpenAI's docs say the agent runs with network access off by default, that locally it uses an OS-enforced sandbox limited to the workspace, and that an approval policy controls when it must stop and ask before acting. Admin controls, SSO and audit logs are listed for Business and Enterprise. These are vendor statements; we did not test them.

What happens to my data?

The pricing page says Business data is not used for training by default and links to OpenAI's business-data page. We did not review the consumer-plan data terms for this entry, so check OpenAI's current privacy terms for the plan you use.

Has AgentComparison tested it?

No. This record is built from the vendor's own documentation, checked 2026-10-06. Vendor claims are attributed, not confirmed.

How to evaluate OpenAI Codex yourself

  1. Pick one low-risk task that matches its documented capabilities: terminal agent that inspects code, edits files and runs commands in your repository.
  2. Write the expected output and what counts as failure before you start.
  3. Connect the minimum access needed, and check approval settings first.
  4. Run it twice and compare, then review every action it took.
  5. Check cost after any free allowance. Unknown price is not free.

Related

Documented use cases: Software development and repository work.

Sources and history

Change history

Documentation overview expanded with current pricing/access sources. No hands-on results claimed.

Report a correction