Get the app
CATALOG

All models

Full catalog of 18,391 indexed models. Filter by family, size, author, license, or data availability. Data refreshed hourly from Hugging Face.

Your rig guess connected · Rough estimate - likely fits models up to ~7 GB. Install the free app for exact numbers + chat with any model locally. - RAM · - free Get the app →
API Rebuilt hourly · same schema

Take the whole catalog via one endpoint.

Filter, page, dump, walk the taxonomy. Every row you see below is machine-readable in one call.

Four shapes cover most use

Full guide · 15 recipes →
GET · filter
A slice of the catalog
/models?family=qwen&method=M4&limit=1000
up to 1,000 rows5–300 cr
GET · one record
One model, full JSON card
/models/{author%2Fname}
46 fields · files · benchmarks10 cr
GET · dump
Everything, page by page
/models?sort=first_seen&limit=1000&offset=N
quadratic offset price5 → 5,100 cr
GET · taxonomy
M-code method breakdown
/methods · /methods/{id}
counts, top authors5 cr
NEW

Change stream — what shifted since your last check.

All seven endpoints →
GET · tombstones
What HF pulled
/models?delisted=dead · /models/delisted
GET · README diff
Narrative shifts
/{id}/drifts · /readme-at/{commit}
GET · trends
Download spikes and drops
/market/anomalies · /download-history
Try it — mirrors this page no filters set — showing catalog top 
curl -H "Authorization: Bearer $ABL_KEY" \
     "https://abliteration.org/api/v1/models?sort=popular&limit=1000"
Get an API key → Pricing → Full-export guide → SDKs · 8 languages →
popular new updated downloads size likes 18,391 results
Catalog is the map. Apps are the tools.

Run models on your own machine, not in the cloud.

Every model page has an "Open in Abliteration" button that hands the model directly to the first-party desktop client, at the quantization your rig can actually run. No API keys, no subscription, no prompt leakage.

Get the app Browse the catalog
Open in Abliteration