How to start with OpenAI Codex in the UAE and Saudi Arabia: plans, setup and first steps
Codex comes with ChatGPT plans, including Free, but what you can do with it depends on your plan and where you run it. Below are setup instructions for each interface, USD prices converted to dirhams and riyals, and details of OpenAI's UAE data residency.
Nikolay Zuykov · Agenticum · 1 November 2026
What Codex is in November 2026 and where it runs
Codex is OpenAI's coding agent, and it is included in ChatGPT plans. You can use it in the ChatGPT desktop app, on the web, in the terminal through the Codex CLI, in a code editor through the IDE extension, and in Codex Cloud for remote work. If a guide you are reading still names GPT-5.5, it is out of date: OpenAI retired GPT-5.5 from Codex with ChatGPT sign-in on 14 October 2026.
| Surface | What it does | Sign-in | Plan notes |
|---|---|---|---|
| Desktop app (macOS, Windows, Linux) | Reads and edits files in a folder you open | ChatGPT or API key | Free and Go get GPT-6 Luna at Standard speed only, subject to rollout |
| Web | Starts Codex Cloud tasks from the browser | ChatGPT | Included in Plus |
| CLI | Runs the agent in your terminal, in scripts and in CI | ChatGPT or API key | Included in Plus; with an API key, billed at API pricing |
| IDE extension | Codex panel inside VS Code, Cursor, Windsurf; separate integrations for JetBrains and Xcode | ChatGPT or API key | Included in Plus; with an API key, billed at API pricing |
| Codex Cloud | Works on your GitHub repositories in a prepared environment, then you commit or open a pull request | ChatGPT only | No cloud features with an API key |
Which plan you need: Free, Go, Plus, Pro, Business, Enterprise or an API key
AED and SAR figures below are conversions of OpenAI's USD prices at the fixed pegs (3.6725 and 3.75).
| Plan | USD / month | AED | SAR | What you get |
|---|---|---|---|---|
| Free | $0 | 0 | 0 | GPT-6 Luna at Standard speed in the desktop app, subject to rollout |
| Go | $8 | 29.38 | 30.00 | Same model access as Free, for lightweight coding tasks |
| Plus | $20 | 73.45 | 75.00 | Codex on the web, in the CLI, in the IDE extension and on iOS |
| Pro | $100 / $200 / $500 | 367.25 / 734.50 / 1,836.25 | 375 / 750 / 1,875 | No five-hour limit for now; Ultrafast mode on the $500 tier |
| Business (annual, 2+ users) | $20 per user | 73.45 | 75.00 | No training on your business data by default |
| Business (monthly) | $25 per user | 91.81 | 93.75 | As above, billed monthly |
| Enterprise & Edu | Not listed here | Data retention and data residency controls; Ultrafast on eligible plans | ||
| API key | API pricing | Local use only; no cloud features such as GitHub code review or Slack | ||
How usage works: for Plus and Standard Business, OpenAI publishes estimated local messages per five-hour window. GPT-6.1 Sol gives roughly 15 to 160 local messages per five hours and GPT-6 Luna roughly 350 to 3,000. OpenAI says these are estimates, not fixed caps. Local messages and cloud chats draw from the same allowance, and weekly limits may also apply. Fast mode uses your included usage at 2.5 times the Standard rate. Plus and Pro users who reach their limit can buy extra credits instead of upgrading.
Steps: use Codex in the ChatGPT desktop app
- Download and install the ChatGPT desktop app for macOS, Windows or Linux.
- Sign in with your ChatGPT account. An API key also works for local work.
- Start a chat, create a project or open a folder. Codex can read and modify files in the folder you choose.
- Select Codex from the ChatGPT dropdown.
- Send a first task. Example only:
Explain what this project does and list the commands that run its tests.
Steps: install and run the Codex CLI
- Install on macOS or Linux with the standalone script. Running it again updates the CLI.
OpenAI's CLI page also lists separate Windows and npm install options. Copy the command for your system from that page.curl -fsSL https://chatgpt.com/codex/install.sh | sh - Make a Git checkpoint. OpenAI recommends one before and after each task so you can revert changes.
git add -A && git commit -m "checkpoint before codex" - Open your project directory and start Codex:
codex - On first launch, choose Sign in with ChatGPT.
- Type your first prompt, let Codex work, then commit a second checkpoint.
Useful next commands:
- API key for CI/CD. OpenAI recommends API-key sign-in for programmatic workflows and warns against exposing Codex execution in untrusted or public environments.
printenv OPENAI_API_KEY | codex login --with-api-key - Non-interactive runs. Call
codex execfrom scripts and pipelines. - Switch models. Type
/modelin a session to pick a model or change reasoning effort. OpenAI positions GPT-6 Astra for the most demanding work and describes GPT-6.1 Sol as offering near-Astra performance at a lower cost. - Local review. The CLI has a review mode for uncommitted changes, a commit or a base branch. It reports prioritized findings without modifying your working tree.
Steps: Codex in VS Code, Cursor, Windsurf, JetBrains and Xcode
- In VS Code, Cursor or Windsurf, install the Codex extension. In JetBrains IDEs or Xcode, use that IDE's own Codex integration.
- In VS Code, Cursor or Windsurf, choose the Codex icon. If you cannot see it, open the Command Palette and run
Codex: Open Codex Sidebar. - Sign in with ChatGPT or an API key.
- Describe your task in the panel.
- For remote work, use Codex Cloud. It requires ChatGPT sign-in.
Steps: run tasks in Codex Cloud with GitHub
- On the web or in the desktop app, choose Work in > Cloud and open Select environment.
- Choose your GitHub repositories and select Get started. Connect GitHub if prompted.
- Codex inspects the repositories, installs dependencies and tools, and tests the workflow.
- Review the setup and test results, save, and select Publish. Wait for Environment published; tasks cannot start from an unpublished environment.
- Start a task. When it finishes, review the changes and test results, ask for follow-ups, then commit or open a pull request.
For reviews inside GitHub, tag @Codex in a pull request or turn on automatic reviews for the repository. This counts as Code Review usage and is not available with API-key sign-in.
Stay in control: sandbox, approvals and AGENTS.md
By default, Codex runs with network access turned off. Locally it works inside an OS-enforced sandbox that usually limits it to the current workspace.
- Auto preset (for example
--sandbox workspace-write --ask-for-approval on-request): Codex reads files, edits them and runs commands in the working directory without asking each time. - Read-only: run
/permissionsand switch to read-only when you only want to chat or plan.
Codex reads AGENTS.md files before it does any work, so put project rules there. By default, Codex loads up to 32 KiB of combined AGENTS.md content, so keep these instructions concise. Example only:
# AGENTS.md
- Run the test suite before proposing a commit.
- Do not edit files under /config/production.UAE and Saudi Arabia: availability and data residency
Both the United Arab Emirates and Saudi Arabia are on OpenAI's list of supported countries. OpenAI says access from countries outside that list may get an account blocked or suspended.
For data location, the two countries differ:
- UAE: ChatGPT data residency (storage at rest) is available in the UAE. Inference residency is also available there, and it requires data residency in the same region.
- Saudi Arabia: not among OpenAI's listed data residency regions.
Checklist before you rely on UAE residency for Codex work:
- OpenAI's pricing page lists data residency controls under Enterprise & Edu. Confirm with OpenAI whether your plan qualifies.
- The Codex app and CLI are supported for residency. Codex Web is not eligible for data or inference residency.
- GPT-6.1 Terra does not support UAE inference residency, so choose another model for that work.
- Business plans do not train on business data by default; Enterprise adds retention controls on top.