Sold by: Prior Labs
Deployed on AWS
TabPFN-2.5 is a Tabular Foundation Model (TFM) used to make predictions on structured data without hyperparameter tuning and with minimal preprocessing. It is a pretrained transformer that provides instant predictions for classification, regression, and time-series tasks across messy tabular data (numeric, text, categorical, missing values, outliers, uninformative features). TabPFN-2.5 handles datasets up to 50,000 samples and 2,000 features and outperforms tree-based models and multi-hour tuned ensembles. This TabPFN-2.5 model package is free to use under the non-commercial conditions as specified in the model license.
Overview
Note: As of June 2026, a new version of this model is available, TabPFN-3-Plus. It is also available on AWS Marketplace.
TabPFN-2.5 by Prior Labs is the worlds leading Tabular Foundation Model. It ranks first on the popular TabArena benchmark for classification and regression tasks, outperforming tree-based models and ensembles tuned for more than 4 hours on datasets up to 50,000 samples and 2,000 features. TabPFN is a pretrained transformer trained on hundreds of millions of synthetic prediction tasks, allowing it to generalize across thousands of use cases in a single forward pass. This enables fast, accurate predictions with minimal preprocessing. The model handles mixed feature types (text, numerical, categorical), missing values, uninformative features, and outliers. It is an ideal default model for teams seeking reliable performance without costly tuning or retraining cycles. In addition to classification, regression, and time-series tasks, TabPFN can be used for unsupervised workflows such as synthetic data generation, uncertainty estimation, and learning tabular embeddings. TabPFN-2.5 is the third generation of the TabPFN models previously published in Nature. This TabPFN-2.5 model package is free to use under the non-commercial conditions as specified in the model license.
Highlights
- Achieve state-of-the-art classification and regression performance in seconds. TabPFN-2.5 removes the need for model selection and hyperparameter tuning, delivering fast and accurate predictions with minimal setup.
- Applies to structured data across any industry including healthcare, finance, manufacturing, energy, and more. Proven across thousands of real-world use cases and diverse tabular datasets.
- No retraining needed. Update the model context with new data and get updated predictions immediately.
Details
Sold by
Delivery method
Latest version
Deployed on AWS
New
Introducing multi-product solutions
You can now purchase comprehensive solutions tailored to use cases and industries.
Features and programs
Financing for AWS Marketplace purchases
AWS Marketplace now accepts line of credit payments through the PNC Vendor Finance program. This program is available to select AWS customers in the US, excluding NV, NC, ND, TN, & VT.
Pricing
This product is available free of charge. Free subscriptions have no end date and may be canceled any time.
Additional AWS infrastructure costs may apply. Use the AWS Pricing Calculator to estimate your infrastructure costs.
Dimensions summary
This model is free to license, so you pay only for the AWS compute hours it runs on. Pricing is charged per host hour and varies by the instance type you select. You choose between two inference modes: batch mode for processing data in groups, and real-time mode for instant predictions. Within each mode, options span CPU-based instances, general GPU instances, and higher-powered GPU instances. Larger instance sizes provide more compute capacity per hour. You match the instance family, size, and mode to your dataset size, speed needs, and workload volume.
Top-of-mind questions for buyers
One HostHr is one hour that an inference instance runs, whether or not it processes predictions. Real-time instances stay running and accrue charges continuously until you shut them down. Batch instances run only while a job processes, so charges stop when the batch job finishes.
Batch mode processes data in groups, so you run an instance only during a job and pay for those hours. Real-time mode keeps an instance running to serve instant predictions, so you pay for every hour it stays live. Batch suits scheduled runs; real-time suits continuous prediction workloads.
The model license carries no software fee. You pay only the AWS compute charge for the instance you select, billed per host hour. Costs scale with the instance family and size you choose and how long each instance runs. Larger instances cost more per hour but provide more compute capacity.
priorlabs.ai+1
Helpful?
Vendor refund policy
Using TabPFN-2.5 on Amazon SageMaker is free of charge for non-commercial use. Because the model itself does not incur any licensing or usage fees, no refunds are provided for any costs incurred while using this product, including but not limited to charges for AWS compute instances, storage, networking, or any other AWS infrastructure used to run or host the model.
If you have questions about this policy or need assistance, you can contact Prior Labs at: hello@priorlabs.ai
Legal
Vendor terms and conditions
Upon subscribing to this product, you must acknowledge and agree to the terms and conditions outlined in the vendor's End User License Agreement (EULA) .
Content disclaimer
Vendors are responsible for their product descriptions and other product content. AWS does not warrant that vendors' product descriptions or other product content are accurate, complete, reliable, current, or error-free.
Delivery details
Amazon SageMaker model
An Amazon SageMaker model package is a pre-trained machine learning model ready to use without additional training. Use the model package to create a model on Amazon SageMaker for real-time inference or batch processing. Amazon SageMaker is a fully managed platform for building, training, and deploying machine learning models at scale.
Deploy the model on Amazon SageMaker AI using the following options:
Version release notes
Fixes an issue for regressor and specifying quantiles
Additional details
Inputs
- Summary
You can submit inference requests to TabPFN-2.5 using two supported input formats:
- application/json - a JSON-encoded request body
- multipart/form-data - with a single Parquet file
Both formats must stay within the 25 MB SageMaker payload limit. Because Parquet is compressed, multipart requests allow more rows or features to fit into the same limit.
Each request may include optional model_params (to configure how the underlying estimator runs) and predict_params (to control the output format of predictions). These parameters follow the same structure in both JSON and multipart inputs.
See examples and fields descriptions below.
- Limitations for input type
- All request payloads, regardless of their content type must be within 25 MB. Larger requests will be automatically rejected by Amazon SageMaker.
{
"task": "classification",
"data": {
"x_train": [
[
1.2,
3.4
],
[
5.6,
7.8
],
[
2.1,
4.3
]
],
"y_train": [
0,
1,
0
],
"x_test": [
[
3,
4.5
],
[
6.1,
8.2
]
]
},
"model_params": {
"n_estimators": 8,
"random_state": 42,
"fit_mode": "fit_preprocessors",
"inference_precision": "auto"
},
"predict_params": {
"output_type": "probas"
}
}
{
"task": "classification",
"data": {
"x_train": [
[
1.2,
3.4
],
[
5.6,
7.8
],
[
2.1,
4.3
]
],
"y_train": [
0,
1,
0
],
"x_test": [
[
3,
4.5
],
[
6.1,
8.2
]
]
},
"model_params": {
"n_estimators": 8,
"random_state": 42,
"fit_mode": "fit_preprocessors",
"inference_precision": "auto"
},
"predict_params": {
"output_type": "probas"
}
}
Input data descriptions
The following table describes supported input data fields for real-time inference and batch transform.
Field name | Description | Constraints | Required |
|---|---|---|---|
task | The inference task. | Supported "classification", "regression". | Yes |
n_estimators | The number of estimators in the TabPFN ensemble, default 8. We aggregate the predictions of `n_estimators` - many forward passes of TabPFN.. | - | No |
categorical_features_indices | The indices of the columns that are suggested to be treated as categorical, default None. If `None`, the model will infer the categorical columns. | - | No |
softmax_temperature | The temperature for the softmax function, default 0.9. This is used to control the confidence of the model's predictions. Lower values make the model's predictions more confident. | Temperature must be greater than 0. | No |
average_before_softmax | Only used if `n_estimators > 1`, default False. Whether to average the predictions of the estimators before applying the softmax function. | - | No |
ignore_pretraining_limits | Whether to ignore the pre-training limits of the model, default False. The TabPFN models have been pre-trained on a specific range of input data. If the input data is outside of this range, the model may not perform well. | - | No |
inference_precision | The precision to use for inference, default "auto". This can dramatically affect the speed and reproducibility of the inference. Higher precision can lead to better reproducibility but at the cost of speed. | - | No |
fit_mode | Determine how the TabPFN model is fitted. The mode determines how the data is preprocessed and cached for inference. This is unique to an in-context learning foundation model like TabPFN, as the "fitting" is technically the forward pass of the model. | - | No |
memory_saving_mode | Enable GPU/CPU memory saving mode, default "auto". This can both avoid out-of-memory errors and improve fit+predict speed by reducing memory pressure. | - | No |
random_state | Controls the randomness of the model. Pass an int for reproducible results and see the scikit-learn glossary for more information. | - | No |
Resources
Vendor resources
Support
Vendor support
For general support or license inquiries reach out to us via hello@priorlabs.ai
AWS infrastructure support
AWS Support is a one-on-one, fast-response support channel that is staffed 24x7x365 with experienced and technical support engineers. The service helps customers of all sizes and technical abilities to successfully utilize the products and features provided by Amazon Web Services.
Similar products
TabPFN-3-Plus by Prior Labs is the latest generation Tabular Foundation Model. A single forward pass tops the public TabArena benchmark for classification and regression, scales to 1M training rows (at 200 features) or 100k rows (at 2000 features), and runs up to 20x faster than TabPFN-2.5. TabPFN-3-Plus also natively handles text features: string-valued columns are accepted directly, without requiring upstream featurization, and are encoded jointly with numeric and categorical features inside the model. Released under the TABPFN-3.0 License v1.0 for research and internal evaluation.
Customer reviews
No customer reviews yet
Be the first to review this product . We've partnered with PeerSpot to gather customer feedback. You can share your experience by writing or recording a review, or scheduling a call with a PeerSpot analyst.