?
dataSources.stats
Reports what is in a corpus and how it was built: document and chunk counts, storage, and the pipeline in effect.
DataSource.stats() → DataSourceStats
Reports what the corpus contains and how it was built: document and chunk counts, storage, when it last ingested, and the embedding model and chunking settings actually in effect. That configuration is the one the documents were built with, not necessarily the platform default, so this is the reliable answer to how a corpus is set up. On a corpus nothing has created yet, exists is false and the rest read zero.
Inspect the corpus
const stats = await Policies.stats(); stats.documentCount; // documents stats.pipeline?.embeddingModelId; // model actually in use