README.md

---
title: Qwen3 Claude Opus
emoji: 🚀
colorFrom: yellow
colorTo: purple
sdk: docker
app_port: 7860
pinned: false
license: apache-2.0
---

# Qwen3 Claude Opus Chat

A minimal, modern chat interface for the Qwen3-4B-Thinking-2507-Claude-4.5-Opus model with:

- **Collapsible thinking blocks** - See the model's reasoning process
- **Web search** - Toggle to enable real-time web search via MCP
- **Streaming responses** - Real-time token streaming
- **Clean UI** - Modern dark theme with smooth animations

## Environment Variables

Set these in HF Spaces Settings > Secrets:

- `OPENAI_API_KEY` - API key for the inference server (required)
- `BASE_URL` - OpenAI-compatible API base URL (default: `https://llama.gptbox.dev/v1`)
- `MODEL_ID` - Model identifier (default: `qwen3-4b-thinking-2507-claude-4.5-opus-distill`)

## Local Development

```bash
npm install
npm run dev
```

## Docker Build

```bash
docker build -t qwen3-chat .
docker run -p 7860:7860 -e OPENAI_API_KEY=your_key qwen3-chat