About
TabPFN (Tabular Prior-data Fitted Network) is a family of foundation models for tabular data developed by Prior Labs. All models in the family share a transformer architecture pre-trained on millions of synthetic datasets and produce predictions in a single forward pass (zero-shot in-context learning over the full training set). Product line: original TabPFN (NeurIPS 2022), TabPFNv2 (Nature 2025, up to 10K rows / 500 features, open-weights), TabPFN-2.5 (up to 50K rows / 2K features), TabPFN-TS (zero-shot time series), TabPFN Enterprise (fine-tuning, real-time, large-scale).
