🔍 VLMs OCR Playground

Compare and use multiple state-of-the-art Vision-Language OCR models:

• DeepSeek-OCR-2: Document to markdown with layout detection
• GLM-OCR: Specialized recognition for text, formulas, and tables
• PaddleOCR-VL-1.5: Full-page document parsing with layout detection

Select a model below to get started!

⚠️ Current Deployment Status:

• 🚀 DeepSeek-OCR-2: Requires GPU - temporarily unavailable on this CPU-only deployment
• 🔮 GLM-OCR: Available ✅
• 📄 PaddleOCR-VL-1.5: Available ✅ (requires API key)

Full-page document parsing with layout detection and element-level recognition.