WireWeek
lift logo

lift

Open-weights vision model that extracts schema-defined JSON from PDFs and images in one pass

About lift

lift is a 9B-parameter open-weights vision model from Datalab that reads PDFs and images and returns JSON matching a schema you supply, handling multi-page documents in one pass. The maker says it scores 90.2% field accuracy on its own 225-document benchmark. The code is Apache 2.0; weights are free for research, personal use and small companies, and a hosted API has a free tier. Out in June 2026.

Key Features

Schema-driven JSON extraction from PDFs and images

Whole-document processing across pages

Local Hugging Face or vLLM inference

Installs with pip install lift-pdf

Reviews

No reviews yet

Be the first to share your experience with lift!

Launch Details

Platform
Pricing
Freemium
Launch Date
October 19, 2026

About the Creator

Automations picked and listed by the Wire Week team. If you build one of them and want to manage the listing yourself, get in touch.

View Full Profile

Related Data and Scraping Launches

Brandfetch

Brandfetch

Logos, colours and fonts for a...

1 votes
OpenDataLoader PDF

OpenDataLoader PDF

Open source PDF parser that tu...

0 votes
OpenParser

OpenParser

Document parsing API that rout...

0 votes

Join 4+ builders launching every week on Wire Week

A curated weekly launch platform where builders get their automations discovered.

Schedule your launch