Most server-side and SDK workflows assume you have paid API keys. This project flips that assumption: it makes a developer's ChatGPT session usable like an OpenAI API endpoint so you can run model calls (including GPT-5.6 variants and GPT Image 2) from local tooling or hosted apps without purchasing separate API credits.
What Sets It Apart
- Account-to-API bridge: runs a local dev proxy that exposes OpenAI-compatible endpoints (/v1/chat/completions, /v1/responses, /v1/models, images) so existing clients and tooling can talk to ChatGPT sessions with minimal changes — so you can reuse existing SDKs and workflows.
- Multiple credential flows: supports local CLI credentials for personal development and a React "Sign in with ChatGPT" flow for web apps — so hosted apps can request user-bound sessions instead of storing API keys.
- Client adapters & SDKs: provides TypeScript/React adapters and an OpenAI-compatible options adapter so you can plug into Vercel AI SDK, the OpenAI JS client, or any client that accepts a baseURL and fetch — reducing integration work.
- Image & streaming support: exposes image generation/edit endpoints and streaming responses through the proxy, enabling multimodal dev scenarios without the official API key.
Who It's For and Trade-offs
Great fit if you want to prototype or build integrations that rely on a ChatGPT account instead of paid API usage, or if you need a quick compatibility layer to reuse OpenAI-oriented code with ChatGPT sessions. It is also useful for local automation, demos, and apps that let users bring their own ChatGPT accounts.
Look elsewhere if you need an official, supported commercial integration (this project is unofficial), if you plan to pool or share credentials across users, or if you require guarantees about rate limits, uptime, or compliance with OpenAI contractual terms. Security and correct handling of user credentials are the developer's responsibility.
