I’ve reviewed TabPFN in the past. https://macinchem.org/2025/02/06/looking-at-tabpfn and I noticed there was a recent update.
TabPFN is a foundation model trained on around 130,000,000 synthetically generated datasets that mimic “real world” tabular data. These datasets sampled dataset size and number of features, both classification and regression tasks, and Gaussian noise was added to mimic real-world complexities. This can then be used to build models for small- to medium-sized datasets with up to 10,000 samples and 500 features and is claimed to be superior to other methods.
TabPFN-2.5 isn’t just about bigger datasets, it’s about changing how you deploy tabular ML in production. For the first time, you can fit once, predict many times with persistent models in the cloud. No more re-fitting for every prediction batch.
More details are here on GitHub
https://docs.priorlabs.ai/quickstart#using-the-rest-api