SAP AI Core

SAP announced on September 15, 2026 that TabPFN-3.5 Plus from Prior Labs, an SAP company, is now available in SAP AI Core. The model is positioned as a tabular foundation model purpose-built for structured business data, enabling predictions on use cases such as cash flow forecasting, payment delays, supplier risk scor SAP describes TabPFN-3.5 Plus as the most accurate and scalable tabular foundation model available today, citing the TabArena and BeyondArena benchmarks as supporting evidence. CTO Philipp Herzig is quoted emphasizing that the combination of TabPFN-3.5 with the SAP-RPT model family delivers accurate predictions from la

SAP Community's technology blog confirms that SAP completed its acquisition of Prior Labs in July 2026, bringing the Prior Labs tabular AI research team into the SAP family. The post states that the TabPFN-3.5-Plus model is generally available directly through SAP AI Core, and that in November 2026, tabular AI orchestr The blog explains that TabPFN is pretrained on vast synthetically generated tabular data, learning to generalize to any prediction problem and using in-context learning to make predictions on unseen datasets without retraining, hyperparameter search, or manual preprocessing. The post positions the model as benchmark-le

SAP announced on July 28, 2026 that SAP-RPT-1.5 is now generally available on the generative AI hub in SAP AI Core. The post, authored by product and topic expert sherene tan and tagged with the managed tags SAP AI Core, SAP Business AI, and AI Foundation, frames the release as the next step after months of customer pi According to the announcement, SAP-RPT-1.5 targets improved prediction accuracy in complex, high-cardinality scenarios and introduces a new explainability layer intended to help teams build greater trust in AI-driven decisions. The blog cites Clariant's automated tax-code determination on vendor invoices—more than ten

SAP published version 2.16.0 of the SAP Cloud SDK for AI on September 18, 2026, with several changes that directly affect SAP AI Core integrations. The release deprecates the sap-rpt-1-small and sap-rpt-1-large model names (functional until retirement on 2026-12-31), removes the default model name on RptClient so calle On the new-feature side, the 2.16.0 release adds an LlmBatchModel type used by LlmBatchModel and LlmBatchRequest.spec.model for IDE autocomplete in batch processing, adds gemini-3.5-flash-lite to the available model list, and schedules mistralai--mistral-large-instruct for retirement on 2026-09-30. For RPT, it adds a c