TabPFN
TabPFN is an open-source tabular foundation model that performs accurate classification and regression on small-to-medium datasets in seconds, trained purely on synthetic data using PyTorch.
At a Glance
About TabPFN
TabPFN is a tabular foundation model developed by Prior Labs that delivers state-of-the-art predictions on structured/tabular data for classification and regression tasks. It is trained purely on synthetic data and leverages a transformer architecture to make accurate predictions on datasets with up to 100,000 rows and 2,000 features. TabPFN supports local GPU inference as well as cloud-based inference via the TabPFN Client, and integrates with a rich ecosystem of extensions for interpretability, hyperparameter optimization, ensembling, and more.
- TabPFNClassifier & TabPFNRegressor: Install via
pip install tabpfn, then call.fit()and.predict()with a scikit-learn-compatible API — no data preprocessing required. - Multiple model versions: Choose between TabPFN-2.5 and TabPFN-2.6 checkpoints, including variants specialized for large features, large samples, or real-data fine-tuning.
- Cloud inference via TabPFN Client: Use the hosted API client for GPU-free inference — ideal for environments without local GPU resources.
- TabPFN Extensions ecosystem: Access advanced utilities including SHAP-based interpretability, outlier detection, synthetic data generation, many-class classification, hybrid Random Forest approaches, and automated HPO.
- Missing value handling: TabPFN natively handles missing values without requiring imputation preprocessing.
- KV Cache for fast prediction: Enable
fit_mode='fit_with_cache'to speed up repeated predictions at the cost of additional memory. - Offline/headless support: Download model weights manually from HuggingFace and configure via environment variables for CI or air-gapped environments.
- Enterprise Edition: Contact Prior Labs for a commercial license with Fast Inference Mode (distillation to MLP/tree ensemble), Large Data Mode (up to 10M rows), and dedicated support.
- No-code UI: Use TabPFN UX at ux.priorlabs.ai for a graphical interface to explore TabPFN capabilities without writing code.
- Python 3.9–3.13 support: Compatible with modern Python versions; GPU (8GB+ VRAM) recommended for optimal performance on larger datasets.
Community Discussions
Be the first to start a conversation about TabPFN
Share your experience with TabPFN, ask questions, or help others learn from your insights.
Pricing
Open Source (Local)
Free open-source local inference via pip install. Use TabPFN on your own hardware with GPU support.
- TabPFNClassifier and TabPFNRegressor
- Local GPU/CPU inference
- Python 3.9–3.13 support
- All open-source model checkpoints
- TabPFN Extensions ecosystem
TabPFN Client (Cloud API)
Free hosted cloud inference via the TabPFN Client for users without a local GPU.
- Cloud-based inference (no GPU required)
- Native text data support
- Simple API client interface
Enterprise Edition
High-throughput production environment with fast inference, large data mode, and commercial support. Contact sales for pricing.
- Fast Inference Mode (distillation to MLP or tree ensemble)
- Large Data Mode (up to 10 million rows)
- Commercial Enterprise License
- Dedicated integration support
- Access to private high-speed inference engines
Capabilities
Key Features
- Tabular classification and regression
- Scikit-learn compatible API
- No data preprocessing required
- Native missing value handling
- GPU and CPU inference
- Cloud-based inference via TabPFN Client
- Multiple model checkpoints (v2.5, v2.6)
- SHAP-based interpretability
- Outlier detection and synthetic data generation
- Automated hyperparameter optimization (HPO)
- Post-hoc ensembling
- Many-class classification support
- Time-series feature support
- KV Cache for fast repeated predictions
- Offline/headless model usage
- Enterprise Edition with Fast Inference and Large Data Mode