<a href="https://www.npmjs.com/package/openai-oauth">NPM</a> | <a href="#legal">Legal</a>

[!NOTE]

What's new in v2?

Add Sign in with ChatGPT to your apps to let users bring their own ChatGPT accounts for AI. Works across free and paid plans. Try it out on Vercel.

And much more:

• Image Generation: Generate and edit images with GPT Image 2
• GPT-5.6 Support: OpenAI OAuth now automatically gets the newest OpenAI models
• Improved CLI: Run in the background with Detached Mode and sign in via the CLI
• Client Adapters: Connect through Vercel AI SDK, the OpenAI client, or any OpenAI-compatible client
• Apache-2.0 License: Use OpenAI OAuth in both open-source and proprietary applications

Quickstart

Dev Proxy

Turn your ChatGPT account into an OpenAI-Compatible API. Learn more

$ npx openai-oauth@latest

OpenAI-compatible endpoint ready at http://127.0.0.1:10531/v1
Use this as your OpenAI base URL. No API key is required.
Available Models: gpt-5.6-terra, gpt-5.6-sol, gpt-image-2, ...

TypeScript SDK

Access your ChatGPT account directly from TypeScript on your machine

npm i @openai-oauth/local @openai-oauth/ai-sdk ai

If you are not signed in to Codex locally, first run npx openai-oauth login.

import { createOpenAIOAuth } from "@openai-oauth/ai-sdk";
import { openaiCredentials } from "@openai-oauth/local";
import { generateText } from "ai";

const openai = createOpenAIOAuth(openaiCredentials());

const result = await generateText({
    model: openai("gpt-5.4-mini"),
    prompt: "Hello!",
});

Works with any OpenAI-compatible client. Learn more

React Component

Let your users sign in with their ChatGPT accounts.

npm i @openai-oauth/react @openai-oauth/ai-sdk ai @ai-sdk/react

Quickstart for Next.js:

// app/page.tsx
"use client";

import { openaiAuthHeaders, SignInWithChatGPT } from "@openai-oauth/react";
import { useCompletion } from "@ai-sdk/react";

export default function Page() {
    const { complete, completion, isLoading } = useCompletion({
        api: "/api/chat",
        streamProtocol: "text",
    });

    return (
        <>
            <SignInWithChatGPT />
            <button
                disabled={isLoading}
                onClick={async () => {
                    await complete("Hello!", {
                        headers: await openaiAuthHeaders(),
                    });
                }}
            >
                Ask
            </button>


{completion}


        </>
    );
}

// app/api/chat/route.ts
import { createOpenAIOAuth } from "@openai-oauth/ai-sdk";
import { openaiCredentials } from "@openai-oauth/react/server";
import { streamText } from "ai";

export async function POST(request: Request) {
    const { prompt } = await request.json();
    const openai = createOpenAIOAuth(openaiCredentials(request));

    const result = streamText({
        model: openai("gpt-5.4-mini"),
        prompt,
    });

    return result.toTextStreamResponse();
}

First-time users will be prompted to install Sign in with ChatGPT for Chrome or Firefox for secure authentication.

Works with any web framework and OpenAI-compatible client. Learn more

Docs

For more information on each of the packages, refer to package-specific README.md.

What is Supported

• Working Endpoints:
• /v1/responses
• /v1/chat/completions
• /v1/models (account-aware by default, or overridden with --models)
• Streaming Responses
• Toolcalls
• Reasoning Traces

See Known Limitations for more information.

openai-oauth CLI

npx openai-oauth

This starts an OpenAI-compatible endpoint (by default at localhost:10531) that is connected to your ChatGPT account.

Press d to keep it running in the background or q to quit. You can also start it in the background directly:

npx openai-oauth --detach
npx openai-oauth status

Follow its logs or stop it from any terminal:

npx openai-oauth logs --follow
npx openai-oauth stop

If you are not signed in, it will ask you to sign in locally. Your credentials will be stored in ~/.codex (the same place codex CLI uses).

You can always directly sign in (without starting the server):

npx openai-oauth login

Login listens on loopback and uses http://localhost:1455/auth/callback, the local callback URL accepted by OpenAI OAuth.

The CLI also supports a few configuration options that generally do not need to be edited.

openai-oauth CLI Flags

Config	CLI flag	Default	Description
Host binding	--host	127.0.0.1	Host interface the local proxy binds to. Non-loopback hosts expose the proxy to your network.
Port	--port	10531	Port the local proxy binds to.
Model allowlist	--models	Account-specific Codex models discovered from ChatGPT	Comma-separated list of model ids exposed by /v1/models. When omitted, the CLI discovers the models your account has access to.
Codex client version	--codex-version	Latest @openai/codex from npm, with a bundled fallback	Override the Codex client version used for model discovery.
Upstream base URL	--base-url	https://chatgpt.com/backend-api/codex	Override the upstream Codex base URL.
OAuth client ID	--oauth-client-id	app_EMoamEEZ73f0CkXaXp7hrann	Override the OAuth client id used for login and refresh.
OAuth token URL	--oauth-token-url	https://auth.openai.com/oauth/token	Override the OAuth token URL used for login and refresh.
Auth file path	--oauth-file	--oauth-file path if provided, otherwise $CODEX_HOME/auth.json or ~/.codex/auth.json	Override where the local OAuth auth file is discovered.
Open browser	--open / --no-open	--open	Open the login URL in a browser during npx openai-oauth login. Use --no-open to print the URL instead.
Login timeout	--login-timeout-ms	300000	How long the login command waits for the OAuth callback, in milliseconds.

SDK Overview

The openai-oauth SDK allows you to integrate ChatGPT login into your local apps and also enable Sign in with ChatGPT for your users.

The SDK is primarily built around two concepts:

• Credential Sources: A way to get a ChatGPT OAuth session
• Such as local authentication, or when a user auths with Sign in with ChatGPT
• Client Adapters: Allows you to actually use the Credential Source
• Such as with Vercel's AI SDK or with the OpenAI client

In general, the SDK will follow this pattern:

import { openaiCredentials } from "@openai-oauth/local";
import { createOpenAIOAuth } from "@openai-oauth/ai-sdk";
import { generateText } from "ai";

// Get credentials from local
const credentials = openaiCredentials();

// Use those credentials to create an AI SDK object
const openai = createOpenAIOAuth(credentials);

// Use AI SDK to run the request
const result = await generateText({
    model: openai("gpt-5.4-mini"),
    prompt: "Hello!",
});

Credential Sources

These allow you to get the OAuth credentials from OpenAI.

@openai-oauth/local

npm i @openai-oauth/local

Use local Codex credentials which live on your machine (normally at ~/.codex).

import { openaiCredentials } from "@openai-oauth/local";

const credentials = openaiCredentials();

This should work out-of-the-box if you're already logged in with Codex, but to log in again, you can run npx openai-oauth login.

You can also point to a specific auth file:

const credentials = openaiCredentials({
    authFilePath: "/path/to/auth.json",
});

@openai-oauth/react

npm i @openai-oauth/react

Use request-bound credentials from the user's browser, with Sign in with ChatGPT.

import { openaiCredentials } from "@openai-oauth/web/server";

const credentials = openaiCredentials(request);

openaiAuthHeaders() returns a plain header object, so it works with both fetch and AI SDK hooks like useCompletion.

In order to actually establish the credentials in the user's browser, you can use openai-oauth's built-in Sign in with ChatGPT SDK, documented below.

For framework neutral usage, see documentation for @openai-oauth/web in packages/web.

How are web credentials stored?

Your OpenAI credentials are by default stored on your device in IndexedDB and encrypted at rest with WebCrypto. Your app server receives request-bound credentials only when the browser sends them with openaiAuthHeaders(), which returns a plain header object.

openai-oauth lets you bring your own credential storage solution if this is not good enough. See documentation for @openai-oauth/web in packages/web for more information.

Client Adapters

These adapters let you use your openai-oauth credentials in any client.

@openai-oauth/ai-sdk

npm i openai @openai-oauth/ai-sdk

Connect openai-oauth to Vercel AI SDK with this provider.

import { createOpenAIOAuth } from "@openai-oauth/ai-sdk";
import { openaiCredentials } from "@openai-oauth/local";
import { generateText } from "ai";

const openai = createOpenAIOAuth(openaiCredentials());

const result = await generateText({
    model: openai("gpt-5.5"),
    prompt: "Reply with exactly: hello",
});

Learn more about how to use Vercel AI SDK. See supported features above.

Migrating from openai-oauth-provider

Vercel AI SDK integration is now independent of your credential source. openai-oauth-provider will soon be deprecated, and the preferred route for using local credentials with the Vercel AI SDK is shown in the example above.

You now import and provide an extra openaiCredentials, either from @openai-oauth/local for local credentials as before or from another credential source.

@openai-oauth/openai-client

npm i openai @openai-oauth/openai-client

OpenAI JavaScript SDK options adapter.

import { createOpenAIOptions } from "@openai-oauth/openai-client";
import { openaiCredentials } from "@openai-oauth/local";
import OpenAI from "openai";

const client = new OpenAI(createOpenAIOptions(openaiCredentials()));

Custom Adapters

openai-oauth also works with any OpenAI-compatible client as long as it takes a custom baseURL and fetch.

import { createOpenAIOAuthTransport } from "@openai-oauth/core";
import { openaiCredentials } from "@openai-oauth/local";

const credentials = openaiCredentials();

const transport = createOpenAIOAuthTransport({
    auth: () => credentials.getSession(),
});

const baseURL = transport.baseURL;
const fetch = transport.fetch;

For example, here's how you would implement the OpenAI JavaScript SDK manually:

const client = new OpenAI({
    apiKey: "openai-oauth",
    baseURL: transport.baseURL,
    fetch: transport.fetch,
});

Image Generation

Generate and edit images with GPT Image 2 using the same ChatGPT credentials and client adapters.

The dev proxy exposes the OpenAI-compatible /v1/images/generations and /v1/images/edits endpoints:

curl http://127.0.0.1:10531/v1/images/generations \
  -H "Content-Type: application/json" \
  -d '{"model":"gpt-image-2","prompt":"A tiny house in a forest"}'

With Vercel AI SDK:

```ts import { createOpenAIOAuth } from "@openai-oauth/ai-sdk"; import { openaiCredentials } from "@openai-oauth/local"; import