priorlabs
A tabular data foundation model with strong predictive performance on messy, complex real-world business tables.
SITE CAPTUREBased on information published on the service's website. Selection and review policy
WHAT IT DOES
What does it do?
TabPFN-3.5 targets real-world tabular prediction tasks involving uncleaned ERP, CRM, and production data. Plus strengthens signal extraction from text-heavy data, Thinking prioritizes accuracy, and Fast runs up to 6 times faster than the base model. The base model and Fast checkpoints are available through the open-source tabpfn package.
KEY FEATURES
Key features
- Predict from high-dimensional, high-cardinality, text-heavy tabular data
- Use the Plus model to strengthen signal extraction from product descriptions, customer reviews, and internal notes
- Use the Thinking model to handle temporal and grouped data
- Run the Fast model up to 6 times faster than the base model
- Access Plus, Fast, and Thinking through API and MCP
- Access the base model and Fast checkpoints through an open-source package