TabPFN: Foundation Model for Tabular Data

Outperform traditional machine learning methods with dramatically faster inference. A transformer-based foundation model designed for supervised classification and regression on tabular data.

50K+
Maximum Rows Supported
2K
Maximum Features Supported
100x
Faster Training & Inference

The Challenges of Tabular ML

Slow Training Times

Traditional ML methods require hours or days to train on large datasets.

Performance Plateaus

Conventional models hit performance ceilings on complex tabular data.

Complex Preprocessing

Requires extensive feature engineering and hyperparameter tuning.

TabPFN: The Solution

Lightning Fast

Achieve state-of-the-art performance in seconds, not hours.

Superior Accuracy

Outperforms XGBoost, Random Forest, and other traditional methods.

Zero-Shot Learning

Minimal preprocessing required - just load and predict.

Powerful Features

Built for modern tabular machine learning with cutting-edge transformer architecture

Transformer Architecture

Advanced transformer-based design that captures complex patterns in tabular data.

Large Dataset Support

Handles up to 50,000 rows and 2,000 features with ease.

Real-Time Performance

Achieve state-of-the-art results in seconds, not hours.

Classification & Regression

Supports both classification and regression tasks seamlessly.

Robust Performance

Consistently outperforms traditional methods across diverse datasets.

Easy Integration

Simple pip installation and scikit-learn compatible API.

Benchmark Results

TabPFN consistently outperforms traditional methods across multiple datasets

Accuracy Comparison

TabPFN
92%
XGBoost
88%
Random Forest
85%
Neural Networks
83%

Training Speed

TabPFN
5 seconds
XGBoost
2 minutes
Random Forest
3 minutes
Neural Networks
5 minutes

Quick Installation

Get started with TabPFN in minutes with our simple installation process

Choose Your Installation Method

Official Installation (Recommended)

pip install tabpfn

From Source

pip install "tabpfn @ git+https://github.com/PriorLabs/TabPFN.git"

Local Development

Requires uv (version 0.10.0 or higher)

git clone https://github.com/PriorLabs/TabPFN.git --depth 1
cd TabPFN
uv sync

Basic Usage

Start using TabPFN with just a few lines of code

Python Example

# Import TabPFN from tabpfn import TabPFNClassifier import numpy as np
# Create TabPFN classifier (default TabPFN-2.6) classifier = TabPFNClassifier()
# Prepare your data X_train = np.random.rand(1000, 20) # 1000 samples, 20 features y_train = np.random.randint(0, 2, 1000) # Binary labels X_test = np.random.rand(100, 20)
# Train and predict (in seconds!) classifier.fit(X_train, y_train) predictions = classifier.predict(X_test)
# Get prediction probabilities probabilities = classifier.predict_proba(X_test) print(probabilities)

Perfect For

TabPFN excels in various real-world applications

Finance

Credit scoring, fraud detection, and risk analysis

Healthcare

Medical diagnosis, patient outcome prediction

E-commerce

Customer segmentation, recommendation systems

Manufacturing

Quality control, predictive maintenance

HR Analytics

Employee retention, performance prediction

ML Research

Rapid prototyping, baseline model comparison

Ready to Revolutionize Your Tabular ML?

Join thousands of data scientists who are already using TabPFN to achieve state-of-the-art results in record time.