SAP aims to displace the more common large language models by releasing its own basic “tabular” model, which the company claims will reduce training requirements for enterprises.
The model, called SAP RPT-1, is a pre-trained model with business and enterprise knowledge out of the box. SAP calls it the Relational Foundation Model, which means it can make predictions based on relational databases even without fine-tuning or additional training.
Walter Sun, SAP’s global head of AI, told VentureBeat in an interview that the value of the new model lies in its ability to perform various enterprise functions, such as predictive analytics, out of the box.
“Everyone knows about language models, and many good models already exist,” Sun said. “But we trained the model on data from business transactions, basically Excel spreadsheets, and so we have a model that can do predictive analytics where the value is that it’s out of the box, meaning you don’t need the specifics of a company for the language model to work.”
Sun said that right out of the gate, RPT-1 can essentially create a business model for enterprises based on data gleaned from SAP’s decades of knowledge. Organizations can also plug models directly into applications without additional fine-tuning.
RPT-1, SAP’s first major family of AI models, will be generally available in “Q4 2025” and deployed through SAP’s AI Foundation. While the RPT-1 is currently available, the company said additional models will be made available soon, including an open-source, state-of-the-art model.
SAP will also release a no-code playground environment for experimenting with the model.
Tabular Model vs LLM
Tabular or relational AI models learn from spreadsheets, unlike LLMs, which learn from text and code. RPT-1 not only understands numbers and the relationships between different cells, but it is also able to provide more structured and accurate answers.
When enterprises decide to use RPT-1, they can add more direction to the model through a little context engineering, because the model is semantically aware and learns based on how it is being used.
SAP researchers first proposed the idea that tabular models could demonstrate semantic awareness and learn from content through a paper Published in JuneIt proposed that ConTextTab introduced context-aware pretraining. It uses semantic cues such as table headers or column types to guide model training, enabling the model to build a relational structure with the data. This is the architecture that works best for tasks requiring models with precise answers, such as financial or enterprise use cases.
RPT models build on the ConTextTab function that lets it learn structured business data from SAP’s knowledge graph, and then add more context through usage.
SAP researchers tested ConTextTab against benchmarks and said it is “competitive” against similar models such as TabPFN and TabIFL.
Development of industry-specific models continues
Many enterprises prefer to enhance general LLMs like GPT-5 or Cloud, to basically retrain the model to answer only questions relevant to their business. However, a shift towards Industry-specific models have begun to take root,
Sun said his experience at a previous company, building a very narrow, highly optimized AI model for sentiment analysis, influenced much of what made RPT-1 different.
“It was a very customized model, a narrow model that took specific responses to specific products but it was not scalable,” Sun said. “When LLM came out, it was a model measuring emotion. But there are use cases we can do that LLM can’t do.”
He said these use cases include predictions, such as determining when a shopper will return to a grocery store, which may also involve numerical analysis along with an understanding of the shopper’s shopping habits. However, some LLMs have begun to integrate into spreadsheets, and AI model providers encourage users to upload similar data for teaching context. Microsoft newly added Copilot CapabilitiesIncluding the ability to work in Excel. anthropic Integrate your cloud Models with Excel, its complement Cloud for Finance ServicesChinese startup Manus also provides data visualization tools Which understands spreadsheets, and ChatGPT can create charts from uploaded spreadsheets and other data sources.
However, SAP notes that it’s more than just reading a spreadsheet; The RPT-1 should stand out among its competitors because it requires less additional information about a business to provide its responses.
