How to run GPT-5.6 Inside Claude Code

The two biggest AI labs just ended up inside the same terminal.

OpenAI’s GPT-5.6 Sol, their strongest coding model, now runs inside Claude Code.

Not a hack. Not a jailbreak.

OpenAI’s own Codex lead is publicly sharing the recipe.

Most people saw this trending, saved the post, and moved on.

I actually set it up. Both ways. And one of the viral versions floating around has a broken command that silently fails.

By the end of this issue, you’ll get my Sol × Claude Code Command Kit -every working command, the fixed alias, and the cost-saver settings in one copy-paste block.

THE AI VAULT IS OPEN

Every guide, kit, template, and prompt library I’ve built - in one place. Organized. Searchable. Growing every week. Here’s how access works:

Paid Members: Get one specific guide related to the newsletter from the Vault every issue. Yours for a week.

Founding Members: Full access to everything inside the Vault. Every guide. Every kit. Every drop. Forever.

We’re 2-3 months away from 10K+ subscribers. The moment we hit that number, we’re launching a private free community - exclusively for AI In Public readers.

---

Last night I saw the setup trending on X and didn’t believe it.

So I opened my terminal and ran both methods back-to-back.

Fifteen minutes later, GPT-5.6 Sol was reviewing code that Claude wrote. Inside Claude Code. On one screen.

That’s when it clicked: this isn’t a party trick. It’s a workflow.

---

Old way vs New way

The old way:

Two terminals. Two tabs. Copy code from Claude, paste into Codex, and copy the critique back.

Ask → Copy → Paste → Lose context → Repeat.

The new way:

Claude writes → Sol critiques → You ship.

One terminal. Zero copy-paste. Two models with different blind spots are checking each other’s work.

---

I have created a complete cheat sheet guide for this setup

Save this and USE it later:

---

The Setup

Everything below happens in your normal Mac Terminal. Nothing runs inside Claude Code until Step 5.

Step 1: Check your setup

Two things must be true before you start.

1. Claude Code is installed:

claude --version

If that fails, install it:

npm install -g @anthropic-ai/claude-code

2. Your ChatGPT plan has GPT-5.6 Sol access.

OpenAI currently lists Sol access in Codex for Plus and higher plans. Free accounts won’t work for this.

---

Step 2: Install the proxy

Paste this into your normal Mac Terminal:

brew install raine/claude-code-proxy/claude-code-proxy

Wait until it finishes. Then confirm the installation:

claude-code-proxy --version

You should see a version number printed on the back. That’s your green light.

This is the current Homebrew command published by the proxy repository. No Homebrew? The repo also ships a one-line install script - it’s in the Command Kit at the end.

---

Step 3: Connect your ChatGPT account

Run:

claude-code-proxy codex auth login

A browser window opens. Then:

1. Sign in with the ChatGPT account that has GPT-5.6 Sol access

2. Approve the authorization

3. Return to Terminal

One thing that matters here: this route uses ChatGPT Plus or Pro sign-in, not an OpenAI API account, no API key. No credits. Your existing subscription.

Confirm that authentication worked:

claude-code-proxy codex auth status

A successful result displays your account and token expiry status.

---

Step 4: Start the proxy

claude-code-proxy serve

This is Terminal 1. Keep it open.

It runs a live monitor showing every request that passes through, which is also how you’ll know everything downstream is working.

---

Everything above is free.

Below the line: the copy-paste launch block that actually fires up gpt-5.6 sol inside Claude Code.

I wasted the afternoon so you don’t have to.