TabPFN-3.5 family
TabPFN-3.5 builds on TabPFN-3 and improves most on non-IID settings, where rows are not independent, features are not clean and data types are mixed. It excels at predictions on tables that are text-rich, have high-cardinality and high-dimensionality features, as well as grouped and temporal data. The family includes four versions of the models:TabPFN-3 family
TabPFN-3 supports numerical and categorical data, missing values, and up to 160 classes natively. The family includes a local base model, Plus for native text features, and Thinking for when compute matters less than accuracy.TabPFN-2.6
TabPFN-2.6 was the default model in TabPFN OSS package versions v7.0.0–v7.1.x. It shares the same architecture and dataset constraints as TabPFN-2.5 with improved benchmark performance.TabPFN-2.5 family
TabPFN-2.5 expanded support to larger datasets than TabPFNv2. The family includes a base model for numerical and categorical data and a hosted Plus model that also handles text.TabPFNv2
TabPFNv2 is the model published in Nature and the first TabPFN to handle real-world messy data — mixed types, missing values, and categorical features automatically.Nature Publication
TabPFN Model License
What does the non-commercial license mean for enterprise use?
What does the non-commercial license mean for enterprise use?
Can enterprises use TabPFN-3.5 or TabPFN-3 for internal use cases?
Can enterprises use TabPFN-3.5 or TabPFN-3 for internal use cases?
What is not allowed in enterprise testing?
What is not allowed in enterprise testing?
Can we use results from testing to inform product features or deployment strategy?
Can we use results from testing to inform product features or deployment strategy?
Can we evaluate TabPFN-3.5 or TabPFN-3 for integration into our product?
Can we evaluate TabPFN-3.5 or TabPFN-3 for integration into our product?
If we want to move from testing to production, what is required?
If we want to move from testing to production, what is required?
Do non-commercial rules also apply to derivative models or fine tuned versions?
Do non-commercial rules also apply to derivative models or fine tuned versions?
What about outputs generated by the model?
What about outputs generated by the model?
If we train a model from scratch without the weights, is that covered by this license?
If we train a model from scratch without the weights, is that covered by this license?
Can we use TabPFN to train a commercial model indirectly?
Can we use TabPFN to train a commercial model indirectly?
Where should we go if we want to deploy TabPFN in production?
Where should we go if we want to deploy TabPFN in production?
What about previous TabPFN releases?
What about previous TabPFN releases?
Who do we contact for commercial licensing?
Who do we contact for commercial licensing?