Quick answer: which should you pick?
| If you want | Pick |
|---|---|
| The cheapest entry plan | Codex. It comes with ChatGPT Free ($0) and Go ($8 a month) in the desktop app, subject to rollout; Claude Code starts with Claude Pro at $20 a month, or $17 billed annually. |
| Open source or local models | Codex. Its CLI is open source under Apache-2.0 and runs open models locally through Ollama; Claude Code isn't open source, and Anthropic doesn't support routing it to non-Claude models. |
| GitHub pull request code review | Codex. Automatic reviews and @codex review come with ChatGPT Plus and up, with no workflow file; Claude Code's workflow-free Code Review is for Team and Enterprise, and on other plans you add the Claude Code GitHub Action to your repository. |
| The deepest configuration and hooks | Claude Code. Its hooks can run shell commands, HTTP endpoints, MCP tools, LLM prompts or subagents; Codex hooks run commands or MCP tools and skip prompt and agent hooks. |
Codex vs Claude Code: price and usage limits
| Codex | Claude Code | |
|---|---|---|
| Entry plan price | $0: included in ChatGPT Free and Go ($8 a month) in the desktop app, subject to rollout. Plus is $20 a month | $20 a month for Claude Pro, or $17 a month billed annually. The Free plan doesn't include Claude Code |
| 5-hour window | Usage is counted over a five-hour window. Pro plans currently have no five-hour limit | A session limit starts with your first message and resets in full five hours later. Verified Ultra accounts have no five-hour limit |
| Weekly cap | Weekly limits may also apply. OpenAI doesn't publish a fixed weekly number | Weekly limits apply across all models and reset at a fixed time each week assigned to your account. Some plans also have a separate weekly limit for Fable |
| Shared allowance | Starting in December, Plus, Pro and Business usage is shared across ChatGPT and Codex and measured in tokens. Instant chat doesn't draw from it | Usage in claude.ai, Claude Code and Claude Desktop counts toward the same limits |
| When you hit the limit | Wait for it to reset, upgrade your plan or buy credits | Wait for it to reset or upgrade, and the new plan's limits apply right away. With usage credits turned on, your work continues on credits instead of stopping |
Neither company publishes its limits in a unit you can compare: how far a plan goes depends on the model, the effort level and how long the conversation gets, and both say they may change limits. Check your remaining usage in ChatGPT Settings > Usage for Codex, and in Settings > Usage on claude.ai for Claude Code.
Hit your limit? Reason keeps going on your other connected accounts or Reason credits.
Use GPT and Claude models side by side in Reason
Open Reason on the web or the Mac app
Sign in at reasonmachines.com, or download the Reason Mac app for macOS from reasonmachines.com/download and sign in there. Both open the same Sessions.
Connect the plans you already pay for
In Settings › Models, choose Sign in with ChatGPT to run OpenAI models on your ChatGPT Plus, Pro or Business plan, and add any other subscriptions or API keys you use. Reason runs on your own accounts first; a Reason plan's credits cover the rest.
Run one task, then ask for a second opinion
Give Reason a task on your repository with one model, then start another Session on the same branch with a model from the other family and ask it to review the diff. Each Session shows which model and account it ran on.
Codex and Claude Code side by side
| Codex | Claude Code | |
|---|---|---|
| Made by | OpenAI | Anthropic |
| License | Codex CLI is open source under Apache-2.0 | Not open source: "All rights reserved", use subject to Anthropic's Commercial Terms |
| Plans that include it | ChatGPT Free, Go, Plus, Pro, Business and Enterprise. On Free and Go it is GPT-6 Luna in the desktop app, subject to rollout | Every paid Claude plan: Pro, Max, Ultra, Team and Enterprise. Not the Free plan |
| Entry paid plan | Plus, $20 a month | Pro, $20 a month, or $17 a month billed annually |
| Higher plans | Pro at $100, $200 or $500 a month; Business $20 per user a month billed annually ($25 monthly) | Max $100 and Ultra $200 a month; Team Standard $25 and Premium $125 per member a month ($20 and $100 billed annually) |
| Usage limits | Five-hour and weekly windows may apply; Pro plans currently have no five-hour limit | A five-hour session limit and weekly limits; verified Ultra accounts have no five-hour limit |
| Check what's left | /status in the CLI, or the usage dashboard at chatgpt.com/codex/settings/usage | /usage in Claude Code, or Settings > Usage on claude.ai |
| Past the limit | Plus and Pro can buy credits; anyone can run local chats with an API key at API rates | Usage credits, Claude Console API credits at API rates, or a bigger plan |
| Surfaces | CLI, IDE extension, desktop app, web (Codex Cloud) and iOS | Terminal, VS Code, JetBrains, desktop app and web |
| Cloud tasks | Codex Cloud on eligible Plus, Pro, Business, Enterprise and Education accounts, not Free or Go | Claude Code on the web at claude.ai/code |
| Local models | codex --oss with Ollama, documented by OpenAI and Ollama | Documented by Ollama; Anthropic doesn't support routing Claude Code to non-Claude models |
| GitHub code review | Codex GitHub app: turn on automatic reviews in Codex settings, or comment @codex review on a pull request. Included from ChatGPT Plus, not with an API key | Claude Code GitHub Actions: a workflow in your repository answers @claude in pull requests and issues and can review each pull request, using an API key or a Pro, Max, Ultra, Team or Enterprise plan. Team and Enterprise also get Code Review with no workflow |
| Instructions file | AGENTS.md: a global file in your Codex home, then one per folder from the repository root down. Codex code review follows it too | CLAUDE.md, at organization, user and project level. From v2.1.277, Claude Code also reads a repository's AGENTS.md when there is no CLAUDE.md |
Prices are in US dollars as the vendors list them, excluding tax, and both companies change plans often: check the linked pages before you buy.
Honest tradeoffs
Choose by the model you prefer on your own code, the plan you already pay for, and the limits you actually hit. Neither tool is better at everything, and a single benchmark score doesn't tell you how either one works on your repository: try the same task in both.
Codex is included in more ChatGPT plans, its CLI source is open, and it runs local open models through --oss. Claude Code needs a paid Claude plan or a Console API account, and its usage is shared with claude.ai and Claude Desktop, so heavy chat use leaves less for coding. Both stop or slow down at their plan's limits; both let you keep going by paying at API rates.
Codex has an /import command that brings Claude Code or Cursor setup, projects and chats into Codex, which helps if you move work from one to the other.
- OpenAI: Codex pricing and usage limits
- OpenAI Help: Using Codex with your ChatGPT plan
- openai/codex on GitHub (Apache-2.0)
- OpenAI: Codex CLI commands
- OpenAI: Custom instructions with AGENTS.md
- Anthropic: Claude plans and pricing
- Anthropic: Use Claude Code with your Pro, Max, or Ultra plan
- Anthropic: How do usage and length limits work?
- Claude Code docs: overview and install
- Claude Code license
- Claude Code docs: CLAUDE.md and AGENTS.md
- Ollama docs: Codex
- Claude Code docs: LLM gateways
Keep going when one runs out
Reason works on your repositories with the model accounts you connect, so when Claude Code or Codex hits its limit you can carry the task on in the Reason web or Mac app: push the branch, tell Reason what is done and what is next, and pick a model that still has usage. If an account Reason runs on hits its own limit, Reason moves the turn to the same model on your other accounts, then to Reason credits, and says so.
Questions
Is Codex better than Claude Code?
Neither is better at everything. They run different models under different plans and limits. Try the same real task in both on your repository, and use the other as a reviewer: in Reason you can run both model families on the same branch.
Is Codex free?
OpenAI includes Codex in ChatGPT Free and Go, where it is GPT-6 Luna at standard speed in the desktop app, subject to rollout. Codex Cloud is not included with Free or Go.
Is Claude Code free?
No. Anthropic includes Claude Code in every paid Claude plan, from Pro at $20 a month, or you can pay per token through a Claude Console account. The Free plan does not include it.
Is Codex cheaper than Claude Code?
To start, yes. Codex is included in ChatGPT Free and Go ($8 a month) in the desktop app, subject to rollout, while Claude Code needs a paid Claude plan, starting with Pro at $20 a month or $17 a month billed annually. At $20 a month, ChatGPT Plus and Claude Pro cost about the same, and how much work each gets you depends on the model and the task.
Which has higher usage limits, Codex or Claude Code?
Neither OpenAI nor Anthropic publishes limits you can compare directly. Both count usage over a five-hour window, Claude adds weekly limits and Codex says weekly limits may also apply, and both raise limits on higher plans: Codex on ChatGPT Pro currently has no five-hour limit, and verified Claude Ultra accounts have none either. Claude Code's limits are shared with claude.ai and Claude Desktop, and from December Codex's are shared with ChatGPT on Plus, Pro and Business. Try your own work on both.
Do Codex and Claude Code share usage limits?
No. Codex counts against your ChatGPT plan and Claude Code against your Claude plan. Claude Code does share its limits with claude.ai and Claude Desktop, and OpenAI says ChatGPT and Codex usage will be shared from December for Plus, Pro and Business.
Can I use GPT and Claude models together?
Yes. In Reason's web or Mac app, connect your ChatGPT plan and other accounts in Settings › Models and pick the model per Session, for example one to write the change and the other to review it.