What it does

This is a Vue-based chat web app that wraps multiple LLM providers—OpenAI, Azure, Claude, Gemini, and Ollama—into a single interface with proper user accounts, admin roles, and rate limiting. It started as a fork of the popular ChatGPT-Web demo, then sprouted a backend with user management, conversation snapshots, and shareable static pages à la ShareGPT.

The interesting bit

The first registered user automatically becomes administrator, which is either admirably trusting or a deployment foot-gun depending on your threat model. Rate limiting is baked in (100 calls per 10 minutes by default), and there’s a full-text search index over conversation snapshots—handy for teams drowning in chat history.

Key highlights

• Multi-model backend: OpenAI/Azure, Claude, Gemini, plus local Ollama hosts
• Built-in rate limiting and user management (first user = admin)
• Shareable conversation pages and snapshot directories with English full-text search
• Prompt management with / shortcut
• Optional auto-generated titles via gemini-2.0-flash; falls back to first 100 chars of prompt
• Text and multimedia file uploads (model-dependent)
• Companion Flutter mobile app in the repo

Caveats

• Documentation and UI mix Chinese and English; some docs are Chinese-only
• Auto-title generation requires specifically configuring gemini-2.0-flash
• Acknowledges it was “written with the help of chatgpt”—not a flaw, but know what you’re getting

Verdict

Good fit for small teams or solo devs who want a self-hosted ChatGPT-like experience with actual access controls. Skip if you need enterprise SSO, granular permissions, or a polished English-only experience.