Skip to content
Anyparser Logo Anyparser Logo Anyparser Docs

How to Choose the Right Model

Anyparser offers a variety of parsing models that cater to different document types and use cases. Choosing the right model is crucial for achieving the best results in terms of accuracy, processing time, and cost-efficiency.

Supported Models

1. Text Model

  • Best for: Plain text documents, Word files, and HTML pages.
  • Use Case: Ideal for extracting clean, unformatted text from structured documents without embedded media or complex layouts.
  • Performance: Fast and efficient, typically resulting in lower processing time.
  • Output: Produces well-structured Markdown or JSON output, preserving headings, paragraphs, and basic formatting.

2. OCR (Optical Character Recognition) Model

  • Best for: Images (JPEG, PNG, TIFF), scanned PDFs, and handwritten documents.
  • Use Case: Use this model when you need to extract text from non-digital documents or images. The OCR model converts printed or handwritten text into machine-readable text.
  • Performance: Slower than the Text Model due to the extra processing required for image recognition.
  • Output: Returns structured text or Markdown with embedded content, such as extracted text from scanned documents or images.

3. VLM (Vision Language Model)

  • Best for: Complex documents, scanned books, invoices, receipts, and any document that combines images, handwriting, and structured text.
  • Use Case: This model is perfect for highly complex, mixed-content files where you need detailed and accurate extraction from images, tables, and unstructured text.
  • Performance: Slower than the Text and OCR models due to its advanced processing, but it offers the highest accuracy.
  • Output: Returns Markdown or JSON with text and images, highly structured with preserved formatting and layout.

4. LAM (Large Audio Model)

  • Best for: Audio files (MP3, WAV, etc.), video files (MP4, AVI, etc.), and podcasts.
  • Use Case: If you need transcriptions of spoken content, such as podcasts, interviews, or webinars, this model provides highly accurate transcription with timestamps.
  • Performance: Requires more processing time and resources due to the nature of audio data, but the transcriptions are highly accurate.
  • Output: Provides a transcribed text output with timestamped paragraphs, ideal for video/audio analysis and retrieval-augmented applications.
  • Coming Soon.

How to Select the Right Model

  1. Consider the Document Type:

    • For simple, well-structured text documents (like Word or PDFs), the Text Model is your best option.
    • If you’re working with scanned documents or images (such as receipts or scanned PDFs), use the OCR Model.
    • For complex layouts or handwritten documents, the VLM Model will provide the most accurate results.
    • For audio or video files, the LAM Model is the most suitable.

  2. Think About Accuracy vs. Speed:

    • If you need fast results, the Text Model is your best bet. It’s the quickest model but may not handle complex documents as well as the other models.
    • For maximum accuracy, especially in challenging scenarios (like images or mixed content), choose VLM or LAM, keeping in mind that processing will take longer.

  3. Cost Considerations:

    • The Text Model is the most cost-effective as it processes quickly with minimal resources.
    • The OCR, VLM, and LAM models are more resource-intensive and slower, but they provide higher accuracy for specialized use cases. These models may incur higher processing costs depending on the amount of data and document complexity.

Key Takeaways:

  • Use the Text Model for speed and efficiency with text-heavy documents.
  • Use the OCR Model for images, scans, and unstructured text.
  • For highly complex documents, use the VLM Model.
  • For transcribing audio or video files, the LAM Model is the best choice.