This project enhances the official Open WebUI by providing deep Chinese localization, advanced model billing and usage statistics, and multi-architecture support. It targets users seeking a refined Chinese experience, precise cost control, and seamless deployment on diverse hardware, especially ARM devices.

How It Works

Built upon Open WebUI, this version integrates novel features like OpenAI's Responses API for streaming thought processes and native Gemini SDK support for parameters and tool usage. It introduces a fine-grained billing system (per-call/token, input/output pricing) and offers optimized ARM64 builds, addressing specific pain points for local AI deployment.

Quick Start & Requirements

• Docker: Recommended for most users. Specific images are provided for x86_64 (latest) and ARM64 (latest-arm64) architectures. Docker Compose is also supported.
• pip: For advanced customization or environments without Docker. Requires Python 3.11+, Node.js 18+, and PM2. Install via pip install open-webui-leon and manage with pm2 start "open-webui serve".
• Prerequisites: Python 3.11+, Node.js 18+, PM2. Ensure correct Docker image architecture is used.

Highlighted Details

• Responses API: Full support for OpenAI's /v1/responses API, enabling real-time streaming of model thought processes.
• Gemini Native SDK: Direct integration with Google's SDK, supporting thinking_budget, native web search, and tool calls.
• Fine-grained Billing: Per-call/token billing with distinct input/output pricing and real-time cost statistics.
• Multi-architecture Support: Optimized ARM64 Docker images for Raspberry Pi, Apple Silicon, etc.
• Deep Chinese Localization: Human-verified UI and error message translations.
• Branch Chat: Create parallel conversation branches from any message.
• Context Control: Visual context truncation and custom context count settings.
• Native Web Search: Models can leverage built-in web search capabilities.
• Function Calling Fallback: Automatic retry mechanism for unsupported models.

Maintenance & Community

The project syncs regularly with the upstream Open WebUI main branch. Feedback is welcomed via GitHub Issues and Discussions. No specific community channels (like Discord/Slack) or contributor details beyond the original author are listed.

Licensing & Compatibility

• License: MIT License.
• Compatibility: Permissive for commercial use and integration with closed-source projects.

Limitations & Caveats

Users on ARM devices must explicitly select the -arm64 Docker image to avoid "exec format error". Pip installations require uninstalling any existing official open-webui package to prevent conflicts.