What it does ChuanhuChat is a Python web UI built on Gradio that talks to pretty much every major LLM API—OpenAI, Azure, Claude, Gemini, DeepSeek, plus local models like LLaMA, ChatGLM, Qwen, and MOSS. It adds file-based Q&A, web search, an AutoGPT-style agent, and even GPT-3.5 fine-tuning, all behind a login wall with per-user history isolation.

The interesting bit The author clearly loathes Gradio’s default look: the 5.0 release brags about “glass morphism effects” and animations that make it “not look like Gradio.” It even adapts to phone notches and installs as a PWA. That’s an unusual amount of frontend care for a Python wrapper project.

Key highlights

• Supports 15+ backends including Midjourney and DALL·E 3 (image gen gets a text UI, naturally)
• Built-in RAG: upload files and query against them
• Auto-naming of chat histories via LLM, with regex search
• Multi-API-key load balancing and custom API host switching
• One-click Hugging Face Space deployment

Caveats

• Some models (XMChat, Midjourney) explicitly don’t support streaming
• The README claims GPT-5 support with specific context lengths, which is… optimistic given current OpenAI offerings; treat this as unverified
• Fine-tuning is limited to GPT-3.5, not newer models

Verdict Worth a spin if you manage multiple LLM keys and want a single dashboard with actual UX polish. Skip it if you need deep programmatic control—this is a power user’s GUI, not an API framework.