prithivMLmods/Gliese-OCR-7B-Post1.0

VISIONPricing:Input $0.4 / Output $0.8Concurrent Unit Cost:1Model Size:7BQuant:FP8Context Size:32kPublished:Sep 10, 2025License:apache-2.0Architecture:Transformer0.0K Open Weights Featherless Exclusive Cold

Gliese-OCR-7B-Post1.0 is a 7 billion parameter vision-language model developed by prithivMLmods, fine-tuned from Camel-Doc-OCR-062825 and built on the Qwen2.5-VL architecture. Optimized for document retrieval, content extraction, and analysis recognition, it excels at understanding complex document structures and visual data representations. This model features enhanced context-aware multimodal extraction and linking, with a 32768 token context length, making it suitable for detailed document analysis and information extraction tasks.

Loading preview...

Gliese-OCR-7B-Post1.0: Advanced Document Analysis Model

Gliese-OCR-7B-Post1.0 is a 7 billion parameter vision-language model, fine-tuned by prithivMLmods from the Camel-Doc-OCR-062825 model, leveraging the Qwen2.5-VL architecture. This model is specifically optimized for advanced document comprehension, focusing on Document Retrieval, Content Extraction, and Analysis Recognition. It has been trained on the Opendoc2-Analysis-Recognition dataset to enhance its capabilities in interpreting complex document layouts and visual information.

Key Capabilities

  • Context-Aware Multimodal Extraction: Understands document context and links multimodal elements within documents.
  • Enhanced Document Retrieval & Content Extraction: Efficiently locates and precisely extracts structured and unstructured content from diverse document formats.
  • Analysis Recognition: Specialized in interpreting analytical content, charts, tables, and visual data representations.
  • Superior Performance: Achieves competitive results on OCR and visual QA benchmarks including DocVQA, MathVista, RealWorldQA, and MTVQA.
  • Video Understanding: Supports detailed comprehension of long-duration videos (up to 20+ minutes) for summarization and Q&A.
  • Visually-Grounded Device Interaction: Enables mobile/robotic device operation through visual inputs and text-based instructions.
  • Improved LaTeX and Markdown Rendering: Shows significant improvements in rendering these formats for OCR tasks.

Good for

  • Context-aware multimodal extraction and linking for complex document structures.
  • High-fidelity document retrieval and content extraction.
  • Analysis recognition of charts, graphs, tables, and visual data.
  • Document-based question answering for educational and enterprise applications.
  • Extraction and LaTeX formatting of mathematical expressions.
  • Retrieval and summarization from long documents and multi-modal inputs.
  • Multilingual document analysis and structured content extraction.
  • Robotic or mobile automation with vision-guided contextual interaction.

Finetunes
2 models