Next-Generation Open-Source AI Workspace with Multimodal Vision, Live Webcam Capture, Deep Research, CoT Reasoning Stream, and Multi-CDN Resilient Artifacts
Open Claude is an extensible, feature-rich, and user-friendly AI workspace inspired by Anthropic's Claude chatbot UI. It connects directly to cloud LLM providers (Google Gemini, Groq, OpenAI) and local offline runners (Ollama, LM Studio) with built-in Deep Research, real-time CoT thinking streams, live webcam photo capture, and interactive sandboxed artifacts.
✨ Key Features • 📸 Showcase • 🤖 Supported Providers • 🚀 Quick Start • ⚙️ Configuration • 🤝 Contributing
Dynamic context greetings ("Burning the midnight oil?"), integrated suggestion chips, live vision capability badges, web search toggle, and instant camera / file upload.
Run complete HTML5 games, Canvas applications, interactive React components, and deep research papers side-by-side with real-time code inspection and execution.
- Live Camera Viewfinder: Tap the camera icon (📷) to open a built-in webcam capture modal with live viewfinder, front/back camera flipping, and shutter review ("Use Photo" / "Retake").
- API-Level Dynamic Vision Detection: No hardcoded lists — Open Claude dynamically inspects provider API manifests and metadata (e.g. Ollama
details.families, Gemini multimodal descriptions, LM Studio architectures). - Intelligent Auto-Switching: When you capture or upload a picture while on a text-only model (e.g., Groq
llama-3.3-70b-versatile), Open Claude automatically switches to the provider's top vision model (llama-3.2-11b-vision-preview/gpt-4o). - Clipboard Paste (
Ctrl+V): Paste images or screenshots directly into the prompt box.
-
Interactive Sandbox: Automatically renders and runs:
- HTML5 & Canvas Games: Full game loops, Web Audio API procedural sound synthesis, and local storage state (e.g. Flappy Bird Deluxe).
- React / JSX Applications: Transpiled on-the-fly via Babel Standalone with support for React Hooks, Lucide Icons, and Recharts.
- SVG Graphics: Visual vector rendering.
- Research Papers: Formal academic format with one-click PDF download.
-
Multi-CDN Resilience: Sequential fallback (unpkg
$\rightarrow$ Cloudflare CDNJS$\rightarrow$ jsDelivr) prevents white screens or failures if any individual CDN is blocked or rate-limited. -
Universal Tag Parsing: Flexibly parses
<antArtifact>tags and code blocks from Claude, Gemini, Groq, OpenAI, and Ollama in any attribute order or format.
- Live Streamed Thoughts: Real-time token-by-token rendering of
<think>and<thinking>tags alongside native API reasoning fields (reasoning_content/reasoning). - DeepSeek R1 on Groq & Ollama: Seamlessly view DeepSeek's step-by-step mathematical and logical reasoning.
- Metrics & Tokens: Displays total reasoning token count and elapsed thinking duration.
- Collapsible Design: Elegant gradient accordion to expand or collapse internal monologue.
- Autonomous Multi-Step Loop: Type
deep research: <topic>to trigger an autonomous research agent that formulates sub-queries, queries real-time web databases, gathers evidence, and writes comprehensive multi-page whitepapers. - Formal PDF Export: Export compiled research reports to PDF with structured abstracts, methodology, analysis, and bibliography.
- Verifiable Citations: Automatic source citation links with verified publisher logos and favicons.
- Tavily API Integration: Search the live web for breaking news, current events, and documentation.
- Smart Query Reformulation: Contextually optimizes search keywords and strips unnecessary prefixes.
- Publisher Logos: Automatically fetches high-resolution publisher favicons for all referenced sources.
| Provider | Active Available Models | Modality | Speed / Notes |
|---|---|---|---|
| Google Gemini | gemini-3.7-flash, gemini-3.5-flash, gemini-2.5-pro, gemini-2.5-flash |
Text + Vision + Audio | Natively multimodal, 2M+ token context |
| Groq Cloud (LPU) | llama-3.3-70b-versatile, llama-3.1-8b-instant, llama-3.2-11b-vision-preview, llama-3.2-90b-vision-preview, deepseek-r1-distill-llama-70b, mixtral-8x7b-32768 |
Text + Vision + Thinking | Ultra-fast throughput (~500–750 tokens/s) |
| OpenAI | gpt-5.6, gpt-5.2, o3-mini, o1, gpt-4o, gpt-4o-mini |
Text + Vision + CoT Reasoning | Frontier reasoning & function calling |
| Ollama (Local) | llama3.3, llama3.2-vision, deepseek-r1, qwen2.5-vl, mistral, gemma2 |
Text + Vision + Reasoning | 100% private, offline, GPU-accelerated |
| LM Studio | Any loaded GGUF model (qwen2.5-vl, minicpm-v, llama3.3) |
Text + Vision | Local inference via localhost:1234 |
# Clone the repository
git clone https://github.com/Damienchakma/Open-claude.git
cd Open-claude
# Install dependencies
npm install
# Start the dev server
npm run devOpen http://localhost:5173 in your browser.
Click the Settings icon (⚙️) in the sidebar or top navigation bar to configure keys:
- Google Gemini: Get free API key at Google AI Studio
- Groq Cloud: Get free API key at Groq Console
- OpenAI: Get API key at OpenAI Platform
- Tavily (Web Search): Get API key at Tavily Dashboard
Note
Privacy First: All API keys and chat histories are stored 100% locally in your browser (localStorage). Your keys are never sent to any third-party intermediary server.
| Shortcut | Action |
|---|---|
Enter |
Send message |
Shift + Enter |
Insert new line in prompt |
Ctrl + V |
Paste image or screenshot directly into prompt |
Space / ↑ |
Jump / flap in game artifacts (e.g. Flappy Bird) |
P |
Pause / resume interactive game artifacts |
- Frontend: React 18, Vite 5, JavaScript (ES2024)
- Styling: Tailwind CSS + Custom Humanist Warm Palette
- Markdown & Code:
react-markdown,remark-gfm,rehype-raw,react-syntax-highlighter - Icons: Lucide React
- PDF Generation:
jspdf,html2canvas - In-Browser Transpiler: Babel Standalone (
@babel/standalone)
Contributions are what make the open-source community such an amazing place to learn, inspire, and create. Any contributions you make are greatly appreciated.
- Fork the Project
- Create your Feature Branch (
git checkout -b feature/AmazingFeature) - Commit your Changes (
git commit -m 'Add some AmazingFeature') - Push to the Branch (
git push origin feature/AmazingFeature) - Open a Pull Request
Distributed under the MIT License. See LICENSE for more information.
Made with ❤️ by Damien Chakma