Recognizers

Text recognition backends for extracting text from detected blocks.

Supported Recognizers

Cloud VLM APIs

• OpenAI: GPT-4 Vision, GPT-4o
• Gemini: Gemini 2.5 Flash, Gemini 2.0 Flash
• OpenRouter: Access to multiple VLMs

Local Models

• PaddleOCR-VL: PaddleOCR-VL-0.9B (NaViT + ERNIE-4.5-0.3B)
• 0.9B parameters
• 109 languages support
• No API costs
• Requires GPU (recommended)

Note

Detailed recognizer API documentation coming soon. See Basic Usage for usage examples.