TabPFN-3.5 Plus Now Available in SAP AI Core for Instant Business Predictions

Key Takeaways

• Tabular AI enters SAP AI Core: SAP has made TabPFN-3.5 Plus, developed by Prior Labs, available to customers through SAP AI Core.
• Predictions without conventional model training: The model uses in-context learning to generate predictions from labelled tabular data without requiring a traditional model-training and tuning process.
• Support for business decisions: Potential applications include cash-flow forecasting, payment-delay prediction, supplier-risk assessment, customer-churn analysis and identifying upselling opportunities.
• Less data preparation: The model is designed to handle mixed data types, missing values and high-cardinality fields, such as product codes and customer identifiers.
• Enterprise deployment: Customers can deploy the model through SAP AI Core and integrate predictions into business processes using structured enterprise data.

SAP Expands Enterprise AI with TabPFN-3.5 Plus

SAP has announced the availability of TabPFN-3.5 Plus in SAP AI Core, giving customers access to an advanced tabular foundation model designed to predict business outcomes from structured data.

Announced on September 15, 2026, the model was developed by Prior Labs, an AI research company acquired by SAP in July 2026. Its availability strengthens SAP’s strategy of integrating specialised AI models into enterprise environments to support faster, data-driven decisions.

Unlike large language models that primarily process and generate language, tabular foundation models are designed to work with structured datasets containing rows, columns, numerical values, categories and other business attributes.

These datasets underpin many enterprise decisions, including financial forecasting, procurement, supply chain planning, customer management and risk assessment. SAP’s integration of TabPFN-3.5 Plus aims to help organisations generate predictions from this data without building and tuning a separate machine-learning model for every use case.

Predictive Analytics Without Traditional Model Training

Traditional machine-learning projects often require data preparation, feature engineering, model selection, training and repeated evaluation. These activities can increase development time and require specialist expertise.

TabPFN-3.5 Plus uses in-context learning to generate predictions from labelled examples supplied with a prediction request. Instead of following the conventional process of training a separate model for each dataset, it uses patterns learned during its development to infer relationships from the provided examples.

The model is designed to work with real-world business data containing missing values, mixed data types and columns with many distinct values. This can reduce the amount of preprocessing and configuration required before prediction.

According to SAP, TabPFN-3.5 Plus performed strongly on external benchmarks, including TabArena and BeyondArena. Such benchmark results provide comparative evidence, although performance in a particular business environment will depend on the dataset, prediction task and evaluation method.

Business Use Cases Across Enterprise Operations

SAP identifies several potential applications for the model across business functions.

1. Cash-Flow Forecasting

Finance teams can use historical transactions, payment behaviour, customer attributes and other relevant variables to predict future cash-flow patterns. These predictions can support liquidity planning and financial decision-making.

2. Payment-Delay Prediction

Businesses can analyse historical invoice and customer data to identify invoices that may be paid late. Finance teams can then prioritise follow-ups and improve working-capital management.

3. Supplier-Risk Assessment

Procurement teams can evaluate structured supplier information, including historical performance and delivery-related data, to help identify suppliers that may require closer monitoring.

4. Customer-Churn Analysis

Customer transaction history, engagement patterns and account attributes can be used to estimate churn risk. Sales and customer-success teams may use these insights to prioritise retention activities.

5. Upselling Opportunities

Organisations can analyse purchasing behaviour and customer characteristics to identify accounts that may be suitable for additional products or services.

These examples illustrate how predictions can become part of existing business workflows rather than remaining isolated analytics outputs. Appropriate validation and business rules are still necessary before organisations use predictions to trigger consequential actions.

Integration Through SAP AI Core

TabPFN-3.5 Plus is available through SAP AI Core’s generative AI hub. Customers can deploy the model and submit classification or regression requests using structured datasets.

Classification can help predict categories, such as whether a payment may be delayed or a supplier may require additional review. Regression can estimate numerical outcomes, including financial values or other measurable business indicators.

SAP’s documentation specifies support for datasets containing up to one million context rows and 2,000 input columns, subject to additional limits on request size, prediction output and task configuration.

The integration provides a managed route for customers to access the model within SAP’s AI infrastructure. Organisations must complete the required SAP AI Core onboarding and use the applicable service plan and deployment configuration.

Prior Labs Strengthens SAP’s Tabular AI Strategy

SAP completed its acquisition of Prior Labs in July 2026, bringing the company’s tabular AI research team into the SAP organisation. SAP had previously announced plans to invest more than €1 billion to develop Prior Labs into a leading research organisation focused on AI for structured business data.

The company is positioning tabular foundation models as a complement to language models and AI agents. While language models can interpret documents and support conversational interactions, specialised tabular models can help estimate outcomes from enterprise datasets.

Combining these capabilities could enable organisations to build workflows that interpret business information, predict potential outcomes and support subsequent decisions within enterprise applications.

Conclusion

The availability of TabPFN-3.5 Plus in SAP AI Core marks another step in SAP’s effort to bring specialised AI models closer to enterprise business data.

By supporting classification and regression without requiring conventional model training for each task, the model aims to simplify predictive analytics across finance, procurement, sales and other business functions.

For SAP customers, the opportunity lies in integrating predictions into existing workflows to improve planning, risk assessment and operational decision-making. The value achieved will depend on data quality, task suitability, deployment design and ongoing validation of model performance.