olmOCR-2-7B-1025 (GGUF)

GGUF quantized version of allenai/olmOCR-2-7B-1025 — Allen AI's state-of-the-art OCR vision-language model, optimized for local inference with llama.cpp and Ollama.

olmOCR excels at extracting structured text from documents, PDFs, images, and handwriting — all running locally on your hardware.

Quick Start

With Ollama

ollama run hf.co/richardyoung/olmOCR-2-7B-1025-GGUF

With llama.cpp

huggingface-cli download richardyoung/olmOCR-2-7B-1025-GGUF \
    --include "*Q8_0*" --local-dir ./models

./llama-cli -m ./models/*Q8_0*.gguf \
    --image document.png \
    -p "Extract all text from this document." \
    -ngl 99

Why This Model?

• Best-in-class OCR: olmOCR outperforms many commercial OCR solutions on academic benchmarks
• Local & private: Process sensitive documents without sending them to cloud APIs
• Structured output: Extracts text with layout awareness — tables, columns, headers
• GGUF format: Runs on consumer hardware with llama.cpp (CPU or GPU)

Use Cases

• Extract text from scanned PDFs and documents
• Digitize handwritten notes
• Process invoices, receipts, and forms locally
• Build privacy-preserving document pipelines

Also Available

• olmOCR MLX 8-bit — Apple Silicon optimized
• olmOCR MLX 6-bit — Smaller footprint on Mac
• olmOCR MLX 4-bit — Minimum RAM on Mac

Other Models by richardyoung

• Abliterated/Uncensored models: Qwen2.5-7B | Qwen3-14B | DeepSeek-R1-32B | Qwen3-8B
• MLX quantizations (Apple Silicon): Kimi-K2 series | olmOCR MLX
• OCR & Vision: olmOCR GGUF
• Healthcare/Medical: Synthea 575K patients dataset | CardioEmbed
• Research: LLM Instruction-Following Evaluation (arxiv:2510.18892)

---

Built & maintained by Richard Young · DeepNeuro