DEV Community

Cover image for How to Use GPT-5.5 for Free with Codex?
Hassann
Hassann

Posted on • Originally published at apidog.com

How to Use GPT-5.5 for Free with Codex?

OpenAI released GPT-5.5 on April 23, 2026, and introduced Codex access to every ChatGPT plan—including Free and Go—for a limited time. The fastest way to try GPT-5.5 for free is to install the Codex CLI, sign in with your ChatGPT account, and start using GPT-5.5 from your terminal—no API key or credit card required.

Try Apidog today

This actionable guide covers Codex CLI installation, authentication, model selection, usage limits, and how to integrate Codex into your dev workflow. For a deeper dive into the model, see What is GPT-5.5. For more free usage options, check the GPT-5.5 for free guide.

TL;DR

  • Codex CLI lets you run GPT-5.5 on your local repo with a 400K context window.
  • All ChatGPT plans (Free, Go, Plus, Pro, Business, Enterprise, Edu) get Codex; Free and Go are limited time.
  • Install: 
  npm install -g @openai/codex
  # or
  brew install codex
Enter fullscreen mode Exit fullscreen mode
  • Authenticate: Use ChatGPT OAuth in a browser, or device-code flow on headless servers.
  • Switch models: 
  /model gpt-5.5
Enter fullscreen mode Exit fullscreen mode

Check quota with /status.

  • Integrate with Apidog: Prototype and test API calls before production.

Why Codex is the Easiest Free Path

OpenAI’s API is paid-only by default—GPT-5.5 on the Responses endpoint is $5/million input tokens and $30/million output tokens after general release. Codex sidesteps this: it wraps the same model in a CLI that authenticates via your ChatGPT account, not an API key. Your ChatGPT plan controls rate limits, but you get the real GPT-5.5 model.

Codex vs API

Install Codex CLI

Codex CLI supports two install methods:

# npm (cross-platform)
npm install -g @openai/codex

# or Homebrew (macOS / Linux)
brew install codex
Enter fullscreen mode Exit fullscreen mode

Verify installation:

codex --version
Enter fullscreen mode Exit fullscreen mode

You should see version 0.28.0 or newer—older versions won’t include GPT-5.5.

Codex CLI Install

Authenticate with a ChatGPT Account

On first run, Codex prompts for authentication.

Browser OAuth (Local Machines) 

codex
Enter fullscreen mode Exit fullscreen mode

A browser window opens; log in with your ChatGPT account. The CLI caches your session for future use.

Device Code Flow (Headless Servers) 

codex login --device-auth
Enter fullscreen mode Exit fullscreen mode

You’ll get a code and URL—open the URL on any device, paste the code, and approve. The CLI on your server completes sign-in automatically.

API Key Fallback

If you prefer to use a paid OpenAI API key:

printenv OPENAI_API_KEY | codex login --with-api-key
Enter fullscreen mode Exit fullscreen mode

This charges your API billing account instead of ChatGPT. Useful for team billing.

Pick GPT-5.5 as the Model

Codex defaults to the “recommended” model for your plan. On paid plans, that’s usually gpt-5.5. Free and Go users may need to switch manually.

During a Session 

/model gpt-5.5
Enter fullscreen mode Exit fullscreen mode

The CLI displays the active model and your rate limits.

Launch Flag 

codex --model gpt-5.5
Enter fullscreen mode Exit fullscreen mode

Check Remaining Quota 

/status
Enter fullscreen mode Exit fullscreen mode

Displays your weekly message budget, context window, and trial expiration for Free/Go plans.

First Session: Practical Example

Codex provides a full-screen terminal UI that reads your repo, runs commands, and edits files. Example workflow:

cd ~/Projects/my-app
codex --model gpt-5.5
Enter fullscreen mode Exit fullscreen mode

Inside Codex, try:

> Read README.md, then open scripts/deploy.sh and summarize what it does in five bullets.
Enter fullscreen mode Exit fullscreen mode

Codex will summarize the script. Next:

> Refactor deploy.sh so it exits on any failed step, and add a dry-run flag. Keep backwards compatibility.
Enter fullscreen mode Exit fullscreen mode

GPT-5.5 proposes a diff; approve to apply changes.

Then run tests:

> Run the deploy test suite and show me the failing case.
Enter fullscreen mode Exit fullscreen mode

Test output streams into the session. If anything fails, ask Codex to fix and loop until tests pass.

This workflow matches the model’s strengths. See OpenAI’s launch post for benchmarks and scores.

What Codex Adds Beyond Raw API Calls

The Codex CLI layers these features over the model:

  • Repo context: Reads your file tree and indexes relevant files without pasting raw content.
  • Command execution with approval: You approve all CLI-suggested commands before execution.
  • Diff previews: All file edits show as unified diffs—accept, reject, or edit before committing.
  • Session persistence: Project sessions persist; pick up where you left off.

Building these features over the API alone is non-trivial; the GPT-5.5 API guide explains the DIY approach.

Rate Limits and Caps per Plan

Caps as of April 23, 2026:

Plan GPT-5.5 access in Codex Weekly cap
Free Yes (limited time) Tight; prototype
Go Yes (limited time, 2× Free) Small
Plus Yes Mid
Pro Yes, highest solo caps High
Business Yes, seat-based High per seat
Enterprise / Edu Yes, contract-based Custom

If you hit your cap, Codex returns a clear error—no silent degradation. Use /status to check usage.

Editor and IDE Integration

Codex authentication is shared across the CLI, VS Code extension, JetBrains plugin, and Codex cloud app. Sign in once via CLI; IDE extensions reuse your credentials.

For Apidog users, a practical workflow:

  1. Prototype a request in Codex CLI (run the GPT-5.5 prompt against this file).
  2. Export the prompt and output into an Apidog collection for team sharing.
  3. Switch from Codex to direct API calls as your contract stabilizes and you move to paid keys.

See Apidog in VS Code for integration details.

Keeping the Workflow Safe on Free and Go

Set these guardrails on day one:

  • Require approval for file writes: In ~/.codex/config.json, set "autoApproveWrites": false. Defaults are safe for Free, but Go plans may auto-apply trivial diffs.
  • Limit workspace scope: Run codex from your project directory. Codex reads from the current directory down—avoid running from ~ or other broad scopes.

OpenAI performed third-party safety reviews and cyber red-teaming for GPT-5.5, but always review diffs before applying.

When to Move Off the Free Path

Free and Go access is time-limited. Plan to upgrade if:

  1. You consistently hit quotas: Upgrade to Plus or Pro for more capacity.
  2. You need direct API access: See the GPT-5.5 API guide.
  3. Team billing is needed: Business or Enterprise plans—see the pricing breakdown.

Underlying model remains the same; only billing and access method change.

FAQ

Does Codex run GPT-5.5 Pro too?

No—Pro isn’t currently exposed in Codex. CLI uses standard GPT-5.5; Pro is available in ChatGPT web/app and (eventually) direct API.

Can I use Codex without a ChatGPT account?

No. You need either a ChatGPT login or an OpenAI API key. Free access requires ChatGPT authentication.

How long will Free and Go access last?

OpenAI says “limited time.” Expect weeks to a few months—upgrade as usage grows.

Does Codex work offline?

No. Every GPT-5.5 operation requires a connection to OpenAI’s servers.

How is Codex different from the ChatGPT web app?

Codex runs in your terminal with local filesystem and repo context, plus shell access. The web app doesn’t have these capabilities.

Top comments (0)