This project is intended for research and educational purposes only. Please use it responsibly and refrain from any commercial use.

WebAI-to-API

A FastAPI server that exposes Google Gemini (via browser cookies) as a local OpenAI-compatible API endpoint. No API key required — it reuses your existing browser session.

Compatible with any tool that supports the OpenAI API format: Open WebUI, Cursor, Continue, custom scripts, etc.

---

Deploy with Docker Compose

1. Create a folder and add two files

docker-compose.yml

services:
  web_ai:
    image: ghcr.io/leolionart/webai-to-api:latest
    container_name: web_ai_server
    restart: always
    ports:
      - "6969:6969"
    environment:
      - PYTHONPATH=/app/src
      - CONFIG_PATH=/app/data/config.conf
    volumes:
      - ./config.conf:/app/data/config.conf
    command: uvicorn app.main:app --host 0.0.0.0 --port 6969 --workers 1 --log-level info

config.conf (leave cookies empty for now)

[Browser]
name = chrome

[AI]
default_ai = gemini
default_model_gemini = gemini-3.0-flash

[Cookies]
gemini_cookie_1psid   =
gemini_cookie_1psidts =

[EnabledAI]
gemini = true

[Proxy]
http_proxy =

2. Start the server

docker compose up -d

3. Open the admin dashboard

Go to http://localhost:6969/admin

From there, paste your Gemini cookies and click Connect — no file editing needed.

---

Getting Gemini cookies

1. Open gemini.google.com and log in
2. Open DevTools (F12) → Network tab → refresh the page → click any request to gemini.google.com
3. Right-click the request → Copy → Copy as cURL
4. Paste the cURL command into the admin dashboard — it extracts the cookies automatically

Or manually: DevTools → Application → Cookies → copy __Secure-1PSID and __Secure-1PSIDTS.

Cookies are saved to config.conf in your folder and auto-rotated in the background — no manual refresh needed.

---

Using the API

The server exposes an OpenAI-compatible endpoint. Point any compatible tool to:

Base URL: http://localhost:6969/v1
API Key:  not-needed

Supported models

Model	Description
gemini-3.0-pro	Most capable (requires Gemini Advanced)
gemini-3.0-flash	Fast, efficient (default)
gemini-3.0-flash-thinking	Extended thinking

Example: curl

curl http://localhost:6969/v1/chat/completions \
  -H "Content-Type: application/json" \
  -d '{
    "model": "gemini-3.0-flash",
    "messages": [{ "role": "user", "content": "Hello!" }]
  }'

Example: OpenAI Python client

from openai import OpenAI

client = OpenAI(
    base_url="http://localhost:6969/v1",
    api_key="not-needed",
)

response = client.chat.completions.create(
    model="gemini-3.0-flash",
    messages=[{"role": "user", "content": "Hello!"}],
)
print(response.choices[0].message.content)

---

Endpoints

Method	Path	Description
GET	/v1/models	List available models
POST	/v1/chat/completions	OpenAI-compatible chat (streaming supported)
POST	/gemini	Stateless single-turn request
POST	/gemini-chat	Stateful multi-turn chat
POST	/translate	Translation (alias for /gemini-chat)
GET	/admin	Admin dashboard
GET	/docs	Swagger UI

---

Common commands

docker compose up -d          # start
docker compose down           # stop
docker compose logs -f        # live logs
docker compose pull && docker compose up -d   # update to latest

---

Star History

License

MIT License