Comparison
Claude Code vs Codex
Claude Code comes with a paid Claude plan. Codex comes with a ChatGPT plan, and ChatGPT's plan table marks Codex as limited on Free and Go. Both can work from a terminal on a repo you already have. This page lines up those published plans. It does not declare a tested winner.
Answer
Neither product is its own subscription. Claude's pricing table marks Claude Code as No on Free and Yes on Pro. Pro is $20 a month, or $17 a month when $200 is billed annually. Codex is included in ChatGPT. Free and Go are limited, Go is $8 a month, and Plus at $20 a month is the card that lists the CLI, the IDE extension, the web, and iOS. vibestacklab has not run the same task on both.
Side by side
| Claude Code | OpenAI Codex | |
|---|---|---|
| What you subscribe to | A Claude plan. The pricing FAQ says Claude Code is included on all paid plans and shares their limits. A Console account bills API tokens instead. | A ChatGPT plan. The Codex pricing page says Codex is included in Free, Go, Plus, Pro, Business, Edu, and Enterprise. An API key bills platform rates instead. |
| Free tier | Free is $0. The pricing table marks Claude Code as No. | Free is $0. ChatGPT's plan table says Codex access is Limited. The Codex card says quick tasks, and names GPT-6 Luna in the desktop app, subject to rollout. |
| Entry paid price | Pro at $20 a month, or $17 a month when $200 is billed annually. Claude Code is included. | Go at $8 a month for lightweight tasks. Plus at $20 a month lists the CLI, IDE extension, web, and iOS. ChatGPT's table marks Codex Limited on Go and Yes on Plus. |
| Heavy individual tier | Max 5x is $100 a month and Max 20x is $200 a month on the Claude Code product page. The plans page says Max starts from $100. | Pro at $100, $200, or $500 a month. Ultrafast mode is listed for the $500 tier. The Codex pricing FAQ says Pro plans currently have no five-hour limit. |
| Team entry | Standard seats at $25 a month, or $20 a month billed annually, minimum two members. US prices, tax excluded. Claude Code is included. | Business standard seats at $25 a month, or $20 a month billed annually, minimum two seats. The help article says they include Codex. Prices may vary by country. |
| Higher team seat | Premium seats at $125 a month, or $100 a month billed annually. The help article says they can be mixed with standard seats. | Premium seats at $125 a month, or $100 a month billed annually. The help article and the business pricing page both say 5x more usage than standard, with no 5-hour limit. |
| Where it runs | The product page leads with the terminal and also lists the web, iOS, Android, GitHub, Slack, VS Code, including Cursor and Devin Desktop, and JetBrains. The FAQ says it runs locally. | Plus lists the web, CLI, IDE extension, and iOS, plus cloud code review and Slack. The auth page says Codex cloud requires ChatGPT sign-in. |
| API key instead of a plan | The product FAQ says a Console account consumes API tokens at standard API pricing. | The pricing page says an API key covers the CLI, SDK, and IDE only. Cloud features such as GitHub review and Slack are not included. |
| Lab result | None. Hands-on test pending. | None. Hands-on test pending. |
Fit by use case
| Use case | Closer public fit | Why, from the docs |
|---|---|---|
| Try a coding agent before paying | OpenAI Codex | ChatGPT Free includes limited Codex. Claude's pricing table marks Claude Code as No on Free. |
| You already pay for Claude | Claude Code | The pricing FAQ says Claude Code is included on paid plans and shares their usage limits. You do not add a second subscription for the agent. |
| You already pay for ChatGPT | OpenAI Codex | Codex is included in the ChatGPT plan. Plus is where the pricing card lists the CLI and the cloud integrations. Free and Go stay on the limited row of ChatGPT's plan table. |
| A measured winner on a shared task | neither | No shared task has been run. A fit column here is a reading of the product pages, not a score. |
Read the two tool pages before treating any cell as current. Prices on this page were checked on 17 October 2026 and each figure links to the vendor page it came from.
Two Anthropic pages do not use the same words for fast mode. The Claude Code product FAQ still says fast mode for Opus 5.5 is in research preview, priced at $8 and $40 per million tokens. The Claude Platform release notes for 15 October 2026 say fast mode for Opus 5.5 is out of research preview and no longer needs an access request. Those notes still say fast mode for Opus 5 and Opus 4.8 remains in research preview.
The Team help article says a standard seat includes 1.25 times Pro's per-session allowance and a premium seat includes 6.25 times Pro's. The pricing FAQ says standard seats give more usage than Pro and premium seats give 5 times more than standard. The same help article says a 200k context window. The pricing page says context is up to 1 million tokens and varies by model, including on Team.
The Codex pricing page still says GPT-5.5 retires from Codex with ChatGPT sign-in on 14 October 2026 at noon PDT, and that API-key sessions are not affected. This check is later than that date, and the sentence is still on the page.
Questions
I'm new. Which one can I open without paying?
Codex. ChatGPT's pricing table says Free includes limited Codex, and the Codex pricing card calls that tier quick tasks. Claude's pricing table marks Claude Code as No on Free. Limited access is not a full cloud setup, and this site has not tried either free path.
Do I buy Claude Code or Codex as its own subscription?
No. Claude Code comes with paid Claude plans. Codex comes with ChatGPT plans. On either side, an API key is a separate bill at that vendor's API rates, and Codex's API key does not include cloud features such as GitHub review or Slack.
What is the cheapest paid plan for each?
Claude Pro is $20 a month, or $17 a month when $200 is billed annually, and it includes Claude Code. ChatGPT Go is $8 a month and the Codex page calls it lightweight. Plus is $20 a month and is the card that lists the CLI, the IDE extension, the web, and iOS. Those figures were checked on 17 October 2026.
Which one won?
Neither. The use-case table says which published plan is closer to a kind of work. It is not a timed comparison and it has no score.