Convert PDFs to Markdown with Local LLMs

Fast, private, and free PDF-to-Markdown conversion using a local LLM (e.g., Gemma 3 via Ollama). No cloud APIs, no tokens, no privacy concerns — just elegant Python.

Article: Convert PDFs to Markdown using Local LLMs — Fast, Private, and Free

---

Installation (Option 1: Using uv)

curl -LsSf https://astral.sh/uv/install.sh | sh
uv venv
uv pip install -r requirements.txt

Then run:

uv run src/pdf2md.py

Installation (Option 2: Using pip)

python -m venv .venv
source .venv/bin/activate
pip install -r requirements.txt

Then run:

python src/pdf2md.py

---

Dependencies

• PyMuPDF (for PDF rendering) — not free for commercial use
• Pillow
• Ollama
• A local model like gemma3:12b or gemma3:4b must be installed and pulled

---

Alternative: Fully Open Source Version

Use pdf2image + Poppler instead of pymupdf.

brew install poppler  # or sudo apt install poppler-utils
pip install pdf2image pillow ollama

Then run:

python src/pdf2md_poppler.py

---

Features

• Handles scanned PDFs via image input
• Converts to clean, structured Markdown
• Works offline with local models

---

Convert pdf to image

pip install pdf2image
brew install poppler    # or in Linux: sudo apt install poppler-utils