aurorax-neo/chat2api

What it solves

It converts the capabilities of the ChatGPT Web interface into a compatible OpenAI-style HTTP API. This allows users to use ChatGPT Web accounts (including Plus and Team plans) within applications that expect a standard OpenAI API format, bypassing the need for official API credits while leveraging web-side features.

How it works

The project acts as a proxy server that translates incoming OpenAI-compatible requests into requests the ChatGPT Web interface can understand. It manages a pool of ChatGPT accounts (via access tokens) and handles authentication through either locally defined API keys or a private prefix system that allows direct passing of real access tokens. It also manages session-specific headers like Chatgpt-Account-Id for Team workspaces and _puid cookies for Plus/Team accounts.

Who it’s for

• Developers who want to integrate ChatGPT Web capabilities into their own apps or CLI tools.
• Users who want to manage multiple ChatGPT accounts in a single API endpoint.
• Those looking to use Function Calling and streaming responses with ChatGPT Web accounts.

Highlights

• OpenAI Compatibility: Supports /v1/chat/completions and /v1/responses endpoints with both JSON and streaming output.
• Function Calling: Full support for tools, tool_choice, and legacy functions parameters, including multi-tool calls and result backfilling.
• Account Pool Management: Built-in support for managing multiple accounts, including Team workspace IDs and Plus/Team PUIDs.
• Flexible Authentication: Supports local sk- keys and a configurable private prefix for direct access token transmission.
• Web Management Interface: Includes an admin panel at /admin to manage API keys, account pools, and proxy settings.