Overview
TabPFN-3.5 by Prior Labs is the world's leading Tabular Foundation Model. It ranks first on the popular TabArena benchmark for classification and regression tasks in a single forward pass, reaching 1858 Elo and outperforming every other tabular foundation model as well as AutoML systems tuned and ensembled for 4 hours, on training sets of up to 1,000,000 rows. TabPFN is a pretrained transformer trained exclusively on synthetic prediction tasks, allowing it to generalize across thousands of use cases without retraining. This enables fast, accurate predictions with minimal preprocessing, while cached inference processes the training rows once and scores new rows in well under a second. The model handles text features, mixed feature types (numerical, categorical), high-cardinality categorical columns, missing values, and wide tables of up to 20,000 features, and natively supports many-class classification. It is an ideal default model for teams seeking reliable performance without costly tuning or retraining cycles. Beyond standard classification and regression, TabPFN-3.5 sets the state of the art on realistic non-i.i.d. data with grouped and temporal splits, on string-valued and multimodal tables, on time-series forecasting, and on relational data. This TabPFN-3.5 model package is free to use under the non-commercial conditions as specified in the TabPFN-3.5 License.
Highlights
- Tops the public TabArena benchmark in a single forward pass. 1858 Elo - ahead of every other tabular foundation model and of AutoML systems tuned and ensembled for 4 hours, with no model selection or hyperparameter tuning.
- Pretrained on synthetic tasks only, so it generalizes without retraining. Handles training sets up to 1,000,000 rows and tables up to 20,000 features; cached inference processes the training rows once and scores new rows in well under a second.
- Native handling of mixed feature types: numerical, categorical (including high-cardinality), and free-text columns, plus missing values and many-class targets. No bespoke preprocessing or upstream featurization required.
Details
Features and programs
Financing for AWS Marketplace purchases
Pricing
Dimensions summary
Top-of-mind questions for buyers
Vendor refund policy
Using TabPFN-3.5 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
How can we make this page better?
Legal
Vendor terms and conditions
Content disclaimer
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.
Version release notes
Initial release of TabPFN-3.5-Plus
Additional details
Inputs
- Summary
TabPFN-3 accepts inference requests in two content types:
- application/json -- inline JSON-encoded request body.
- multipart/form-data -- with X_train, y_train, X_test as separate Parquet or CSV file parts plus a JSON config part.
- Limitations for input type
- Sync (real-time) invocations are capped by SageMaker at 6 MB payload and 60 s processing. Async invocations support up to 1 GB payload and 60 min processing, which is the recommended deployment for any non-trivial tabular dataset.
- Input MIME type
- application/json
Input data descriptions
The following table describes supported input data fields for real-time inference and batch transform.
Field name | Description | Constraints | Required |
|---|---|---|---|
task | Inference task type. Required; accepted at top level or nested under `task_config.task`. | AllowedValues: classification, regression. | Yes |
n_estimators | Number of TabPFN forward passes that are averaged for the final prediction. Default is the underlying TabPFN library default. | MinValue: 1. MaxValue: 8. | No |
categorical_features_indices | Indices of columns to treat as categorical. If omitted or null, the server infers categorical columns from the data. | - | No |
softmax_temperature | Temperature for the softmax over class logits (classification) or the predictive distribution (regression). | MinValue: greater than 0. Default: 1.0. | No |
average_before_softmax | Whether to average ensemble logits before applying softmax (instead of averaging post-softmax probabilities). Only meaningful when n_estimators > 1. | Boolean. Default: false. | No |
ignore_pretraining_limits | If true, the server does not reject inputs that exceed TabPFN-3.5's validated row, feature or cell-budget envelope. Outside-envelope inputs may run but with degraded accuracy. | - | No |
inference_precision | Numeric precision used for the forward pass. "auto" selects per-instance defaults. | Default: auto. | No |
fit_mode | Controls how the server caches between the fit-time forward pass and any subsequent predicts. `fit_preprocessors` (default) runs everything per request. `fit_with_cache` caches the fit and returns a model_id which can be reused on later requests via `context.model_id` to skip the fit step. | AllowedValues: fit_preprocessors, fit_with_cache. Default: fit_preprocessors. | No |
memory_saving_mode | If true, reduces GPU/CPU memory pressure during inference. Trades a small amount of throughput for the ability to handle larger contexts. | Boolean or null. Default: null. | No |
random_state | Seeds preprocessing-side randomness for reproducibility across runs. | Integer or null. Default: null. | No |
Support
Vendor support
Support can be reached 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
