WebAI-to-API

WebAI-to-API is a browser-native AI runtime that exposes browser-based AI services through OpenAI-compatible APIs.

---

Features

• OpenAI-compatible /v1/chat/completions API
• Provider-based architecture with unified routing
• Streaming response support (SSE)
• Conversation continuation support
• Health, readiness, and runtime diagnostics endpoints
• Docker deployment support
• Authentication management and browser login workflows

---

Available Providers

Gemini

Provides access to Google Gemini models through either the WebAPI backend or a browser-native Playwright runtime.

Atlas Cloud

Provides access to cloud-hosted AI models through a native API integration powered by Atlas Cloud.

---

Quick Start

Prerequisites: Git, Python >=3.11,<3.13 and Poetry. On Windows, use Python 3.11.10+ or 3.12.4+ for secure Gemini WebAPI temporary-cookie-cache handling. See the Installation Guide for full installation and troubleshooting details.

1. Install and Set Up

Clone the repository, enter the project directory, then run the setup wrapper for your platform.

Linux / macOS

git clone https://github.com/Amm1rr/WebAI-to-API.git
cd WebAI-to-API
./install.sh

Windows PowerShell

git clone https://github.com/Amm1rr/WebAI-to-API.git
cd WebAI-to-API
.\install.ps1

The wrappers create missing configuration and runtime state, install project dependencies and Playwright Chromium, then run diagnostics. See the Installation Guide for manual setup, troubleshooting, and Make shortcuts.

2. Configure

Review the generated config.conf. Core Gemini settings include:

[Gemini]
backend = webapi
default_model = gemini-3-flash
extended_thinking = false

Configure Atlas separately when needed. See the Configuration Guide for provider, proxy, logging, and authentication settings.

3. Authenticate

For browser-based Gemini authentication:

poetry run python verify_login.py

Gemini WebAPI can also use configured cookies. See the Configuration Guide for authentication methods. For Docker authentication, see the Docker Deployment Guide.

4. Start the Server

poetry run python src/run.py

• API: http://localhost:6969
• Dashboard: http://localhost:6969/ui
• Swagger UI: http://localhost:6969/docs

---

Updating

Host

Linux / macOS

./update-linux-macos.sh

Windows

update-windows.cmd

Updates are version-driven from origin/master. See the Updater Guide for locking, preflight checks, rollback, dependency sync, and platform details.

Docker

git pull
APP_UID=$(id -u) APP_GID=$(id -g) docker compose up -d --build

See the Docker Deployment Guide for Docker setup and deployment details.

---

Send Your First Request

curl -X POST http://localhost:6969/v1/chat/completions \
  -H "Content-Type: application/json" \
  -d '{
    "model": "gemini-3-flash",
    "messages": [
      {
        "role": "user",
        "content": "Hello!"
      }
    ]
  }'

---

Dashboard

Open the dashboard at http://localhost:6969/ui. It provides runtime status, authentication view, model and API discovery, a playground, and conversation management where supported. See the Dashboard Guide.

---

Main Endpoints

Endpoint	Purpose
/v1/chat/completions	Main OpenAI-compatible chat endpoint
/v1/temporary/chat/completions	Temporary Gemini WebAPI chat without durable conversation persistence
/v1/models	Current runtime model catalog
/v1/conversations	Manage persisted Gemini WebAPI conversations
/v1/auth/status	Authentication status
/v1/auth/login	Interactive browser login trigger
/v1/runtime/status	Runtime diagnostics
/health	Liveness
/ready	Runtime readiness
/translate	Translate It! compatibility endpoint
/ui	Dashboard

See API Documentation for the complete API surface, including compatibility and legacy endpoints.

---

Supported Models and Routing

Available models depend on configured providers and runtime availability. Use /v1/models as the authoritative current catalog.

gemini-3-flash
playwright/gemini-3-flash
atlas/<model-id>

Unprefixed Gemini models use the configured Gemini backend. playwright/... forces browser-native Gemini routing, while atlas/... routes to Atlas. See API Documentation for full routing behavior.

---

Configuration Summary

Configure Gemini backend selection (webapi or playwright), default model, provider enablement, proxy, logging, and Atlas API access in config.conf and .env. Set a default for Extended Thinking with [Gemini].extended_thinking, or override it per request with provider_options.gemini.extended_thinking. See the Configuration Guide.

---

File Support

OpenAI-style file content parts are supported by Gemini WebAPI. Gemini Playwright and Atlas do not currently support file parts, and Gemini WebAPI does not preserve exact text/file interleaving. See API Documentation for supported formats and limits.

---

Security

WebAI-to-API does not provide caller API authentication. Keep the default localhost binding unless external authentication and access control protect the service. See the Docker Deployment Guide and Dashboard Guide.

---

Documentation

• Installation Guide
• API Documentation
• Configuration Guide
• Architecture Guide
• Docker Deployment Guide
• Dashboard Guide
• Updater Guide

Interactive API documentation is available through Swagger UI when the server is running.

---

Star History

---

License

WebAI-to-API is licensed under the Apache License 2.0. See LICENSE for the full text.