SAP | AI Tools Showcase
Tabular AI · RPT-1 & TabPFN

Prediction
Without Training

Upload a CSV, pick a target column, and get predictions from SAP's pretrained tabular model — no model training, no code required.

Applications

RPT-1

RPT-1 Predictor

Upload one or more CSVs and predict missing values using SAP RPT-1.

Open app →
Benchmark

RPT-1 vs TabPFN

Run both models on the same dataset and compare predictions side-by-side.

Open app →

What is RPT-1?

RPT-1 (Relational Pretrained Transformer) is SAP's foundation model for tabular data. It uses in-context learning — your labeled rows become the model's few-shot examples, and it predicts unknown values in a single API call. No training pipeline, no feature engineering.

It handles both classification (categories, statuses, labels) and regression (numeric targets like price or quantity) from raw tabular data.

Core idea

In-Context Learning

Labeled rows act as few-shot examples. No retraining or fine-tuning.

Data

Any Tabular Data

Mixed numeric and categorical columns work out of the box.

Output

Confidence Scores

Each classification prediction includes a confidence percentage.

Setup

API Key Only

Just a CSV and an API key. No infrastructure or pipelines needed.

API keys required. The RPT-1 Predictor requires an RPT-1 API key from rpt.cloud.sap. The comparison app also requires a TabPFN access token from tabpfn.com. Keys are sent per-request and never stored server-side.