HuggingFace GGUF Model Tracker

Download and keep a local CSV of all GGUF models on HuggingFace, updated incrementally.

---

What's in this folder

File	What it does
fetch_all.py	First-run script — downloads ALL ~156k GGUF models to CSV
update.py	Daily/weekly updater — only fetches models newer than your last run
updater.html	Browser-based GUI — drag in CSV, click sync, download updated file
huggingface_gguf_models.csv	Your local database (not included — run fetch_all.py first)

---

Quick start

Option A — Python scripts (recommended)

Requirements: Python 3.7+ and the requests library.

pip install requests

Step 1 — First time only: download everything (~10 min, ~51 MB)

python fetch_all.py

This creates huggingface_gguf_models.csv with all GGUF models.

Step 2 — Every time after: just fetch new models (seconds to minutes)

python update.py

The script reads your existing CSV, finds the latest Created At date, fetches only models created after that date, and prepends them to the file. Run this daily, weekly, or whenever you want fresh data.

---

Option B — Browser GUI (no Python needed)

1. Open updater.html in any browser (double-click it)
2. Drop your huggingface_gguf_models.csv onto the page
3. Click "Fetch new models" — it syncs directly from the HuggingFace API
4. Click "Download updated CSV" — saves the merged file

Note: The browser version fetches from the HuggingFace API directly. No server, no backend — it runs entirely in your browser.

---

CSV columns

Column	Description
Model ID	Full author/model-name (e.g. unsloth/Qwen3-8B-GGUF)
Author	Uploader / organization
Model Name	Just the model name part
Pipeline / Task	e.g. text-generation, image-text-to-text
Likes	Heart count
Downloads (30d)	Downloads in the last 30 days
Trending Score	Current trending rank score
Last Modified	Date of last file change
Created At	Date the model was first uploaded
Private	Yes / No
Gated	Yes / No (requires approval to access)
Tags	Up to 20 tags, pipe-separated
GGUF File Count	Number of .gguf files in the repo
URL	Direct link to the model page

---

Tips

• First run takes ~10 minutes — HuggingFace rate-limits at ~5 req/sec
• Subsequent updates are fast — only new models are fetched
• The CSV is sorted newest-first after each update
• Duplicate detection is by Model ID — no duplicates even if runs overlap
• You can open the CSV in Excel, LibreOffice, DBeaver, or any tool

---

Custom output path

python fetch_all.py --output /path/to/my_models.csv
python update.py --csv /path/to/my_models.csv