FastGPT Proxy: Project Overview

1. What is FastGPT Proxy?

FastGPT Proxy is a lightweight proxy allowing FastGPT API to be OpenAI-compatible. This allows FastGPT(ChatGPT 4o varient) to be used as a model in services like Open WebUI which provides STT and TTS, letting you use the 'call' and 'record to send' functions (this is currently unvailable on default FastGPT portal).

2. Technical Stack

Component	Technology	Purpose
Core Language	Python 3.11	Backend logic and API handling
Web Framework	Flask 3.0	HTTP request routing and response
Production WSGI	Gunicorn 21.2	Scalable server for production use
Dependencies	requests, flask-cors, python-dotenv	HTTP calls, CORS, environment variables
Deployment	Docker, GitHub Container Registry (GHCR)	Containerization and distribution

3. Key Objectives

✅ OpenAI API Compatibility : Kagi FastGPT responses into OpenAI’s /v1/chat/completions format for drop-in replacement.

✅ Self-Hosted Alternative : Lets you use FastGPT with TTS and STT functionality provided by Open WebUI including open calls.

✅ Simplified Integration : Single Docker image with minimal setup steps.

4. Limitations

Functional Constraints

• No Streaming Support : Responses are delivered as a single block (no real-time chunks).
• Basic Error Handling : Errors from Kagi API may not be fully translated to OpenAI format.
• Optional Parameter Support : Not all OpenAI parameters (e.g., temperature, top_p) etc.

5. Future Enhancements

• If Kagi decides to offer the models they already offer in with their Kagi Assistant, I will develop a model pull script like what OpenWeb UI provides for OpenRouter: see https://openwebui.com/t/yikesawjeez/yikes_openrouter
• Streaming Support : SSE (Server-Sent Events) for real-time text generation.

7. Quick Start (TL;DR)

Deploy with Docker

docker run -d --name fastgpt-proxy -p 5000:5000 \
  -e KAGI_API_KEY=your_kagi_api_key_here \
  -e API_KEY=your_optional_api_key_here \
  ghcr.io/nomadxxxx/fastgpt-proxy:latest

Then configure Open WebUI to use http://localhost:5000/v1 as the OpenAI endpoint.

The harder way

Clone repository:

git clone https://github.com/nomadxxxx/fastgpt-proxy.git
cd fastgpt-proxy

Create virtual environment:

python3 -m venv venv
source venv/bin/activate  # Linux/Mac | venv\Scripts\activate for Windows

Install dependencies

pip install -r requirements.txt

Configure .env file

.env

nano .env # Edit KAGI_API_KEY and optional API_KEY

Start server

python3 app/server.py

Note you can also you: gunicorn --bind 0.0.0.0:5000 app.server:app