prithivMLmods/Camel-Doc-OCR-080125

VISIONPricing:Input $0.4 / Output $0.8Concurrent Unit Cost:1Model Size:7BQuant:FP8Context Size:32kPublished:Jul 25, 2025License:apache-2.0Architecture:Transformer0.0K Open Weights Featherless Exclusive Cold

The prithivMLmods/Camel-Doc-OCR-080125 is a 7 billion parameter multimodal model, fine-tuned from Qwen2.5-VL-7B-Instruct, specifically optimized for document retrieval, content extraction, and analysis recognition. It excels at understanding complex document structures, extracting both structured and unstructured content, and interpreting analytical data like charts and tables. This model's primary differentiator is its enhanced context-aware multimodal extraction and linking capabilities for documents, making it highly effective for document-based question answering and information extraction tasks.

Loading preview...

Camel-Doc-OCR-080125: Document-Optimized Multimodal AI

The Camel-Doc-OCR-080125 model, developed by prithivMLmods, is a 7 billion parameter multimodal AI based on the Qwen2.5-VL-7B-Instruct architecture. It has been specifically fine-tuned on the Opendoc2-Analysis-Recognition dataset to enhance its capabilities in document retrieval, content extraction, and analysis recognition. This model is designed to efficiently process and understand complex document structures, making it a powerful tool for information extraction and document-based question answering.

Key Capabilities

  • Context-Aware Multimodal Extraction: Advanced understanding and linking of multimodal elements within documents.
  • Enhanced Document Retrieval: Efficiently locates and extracts relevant information from diverse document formats.
  • Superior Content Extraction: Precise extraction of structured and unstructured content.
  • Analysis Recognition: Specialized in interpreting analytical content, charts, tables, and visual data representations.
  • Competitive Performance: Achieves strong results on OCR and visual QA benchmarks such as DocVQA, MathVista, RealWorldQA, and MTVQA.
  • Video Understanding: Supports detailed comprehension of long-duration videos (up to 20+ minutes) for summarization and QA.
  • Visually-Grounded Device Interaction: Enables robotic or mobile device operation via visual inputs and text instructions.

Good For

  • Document-based Question Answering: Ideal for educational and enterprise applications requiring information from documents.
  • High-Fidelity Content Extraction: Extracting data from various document formats, including complex layouts.
  • Analysis of Visual Data: Interpreting charts, graphs, and tables within documents.
  • Multilingual Document Analysis: Structured content extraction for global use cases.
  • Robotic/Mobile Automation: Vision-guided contextual interaction for device operation.

Finetunes
2 models