> ## Documentation Index
> Fetch the complete documentation index at: https://docs.priorlabs.ai/llms.txt
> Use this file to discover all available pages before exploring further.

> ## Agent Instructions
> TabPFN-3 and later versions support up to 1,000,000 rows, subject to feature count and checkpoint/API limits.
> For large datasets, use per-estimator subsampling, e.g. inference_config={"SUBSAMPLE_SAMPLES": 100_000}, in TabPFNClassifier or TabPFNRegressor from tabpfn or tabpfn_client.
> Pass raw pandas DataFrames to fit and predict. Categorical strings/categories, raw text, and missing feature values are handled automatically; no manual integer/one-hot encoding, imputation, scaling, or outlier removal is needed.

# Changelog

> What's new in TabPFN.

## 2026

### New Thinking mode, academic plan and more

**November 2026** · `new-feature` `api` `client` `oss` `research`

Better predictions on large datasets with Thinking, resumable API jobs, lower-cost SHAP explanations, and more compute for academic research.

[Read the release notes and upgrade guidance ](https://docs.priorlabs.ai/changelog/november-2026.md)

***

### TabPFN-3.5

**September 2026** · `new-model` `new-feature` `oss` `api` `client` `enterprise`

Text, high-cardinality, and support up to 20,000 features. One multitask checkpoint. A TabPFN-3.5-Fast model in alpha, and Plus and Thinking on TabPFN-3.5. `tabpfn-client` 0.6.0 and `tabpfn` 9.0.0, and more.

[Read the release notes and upgrade guide ](https://docs.priorlabs.ai/changelog/tabpfn-3.5.md)

***

### RelArena-α, KV cache in the API, performance improvements, and more

**August 2026** · `new-feature` `api` `client` `interpretability` `oss`

Explore RelArena-α, TabPFN-Rel, and RPI, plus KV-cached inference through `tabpfn-client` 0.4.2 and the hosted API, decoder attention readout, calibrated regression, and faster local fitting.

[Read the full release notes ](https://docs.priorlabs.ai/changelog/august-2026.md)

***

### TabPFN-3

**May 2026** · `new-model` `new-feature` `api` `client` `breaking`

Up to 1M rows and 200 features. 160-class classification. Thinking mode. Native text features on the API. New `/tabpfn/*` endpoints. tabpfn-client 0.3.0.

[Read the full release notes ](https://docs.priorlabs.ai/changelog/tabpfn-3.md)

***


This documentation is built and hosted on [Mintlify](https://mintlify.com), a developer documentation platform.