vlmrun-hub

What it is and what it does

VLM Run Hub is a library of pre-built Pydantic data models designed to standardize the extraction of structured information from visual content—images, videos, and documents. Rather than writing custom parsing logic for each extraction task, you import a schema from the hub (e.g., Invoice, DriverLicense, MedicalInsuranceCard) and pass it to a Vision Language Model's structured output API. The model returns data conforming to that schema, which Pydantic automatically validates.

The library covers real-world domains: document processing (invoices, receipts, passports), accounting and finance (W2 forms, balance sheets), healthcare (insurance cards, pathology reports), retail (product catalogs), media (sports game states), and logistics. Each schema has been tested across industry use cases. The schemas are model-agnostic—use the same Invoice schema with OpenAI's GPT-4o, Anthropic's Claude Vision, or the VLM Run SDK without rewriting prompts.

Use it for:

• Extract invoice line items, dates, and customer details from receipt images for accounting automation.
• Parse driver's license or passport images into structured fields for identity verification workflows.
• Convert medical insurance cards into validated data for claims processing systems.
• Extract product information and pricing from retail catalogs or ecommerce images.
• Automate document classification and field extraction for document management systems.
• Build visual ETL pipelines that feed validated structured data directly into downstream databases or APIs.

Worth the install?

AI-flagged interpretation of the facts on this page — verify before relying

Provides a catalog of pre-defined Pydantic schemas for extracting structured data from images, videos, and documents using Vision Language Models.

Yes, if you are building applications that extract structured data from images or documents using Vision Language Models. The pre-built schemas save significant development time and eliminate the need to design and validate extraction formats for common document types. The library is permissively licensed, has low install friction, and covers a broad range of real-world domains. Maintenance is aging but not stalled; evaluate whether the schema catalog meets your specific use cases before committing.

Install

vlmrun-hub on PyPI

pip install vlmrun-hub

uv add vlmrun-hub

poetry add vlmrun-hub

Installing vlmrun-hub

Before you install

Low install friction with a pure-Python wheel. Maintenance status is aging—last commit was 2025-12-15 and the project has 554 stars, suggesting active but not rapid development.

License in practice

Apache-2.0 permissive license allows commercial and private use with minimal restrictions; you must include a copy of the license and state significant changes.

Quickstart

pip install vlmrun-hub

from vlmrun.hub.schemas.document.invoice import Invoice
from pydantic import ValidationError

# Use with instructor + OpenAI or another VLM provider
# Schema validates extracted data automatically

Requires Python 3.9 or later; intended for use with a Vision Language Model provider (e.g., OpenAI, Anthropic, or VLM Run SDK).

Verify before relying

• Whether the catalog of schemas is actively maintained and updated with new domains beyond those listed in the excerpt.
• Performance characteristics when validating large batches of extracted data or deeply nested schemas.
• Compatibility guarantees with specific VLM provider APIs (e.g., OpenAI Structured Outputs, Gemini, Anthropic).

Package facts

Evidence: vlmrun_hub-0.1.35-py3-none-any.whl

Tags

More Libraries packages