Vector storage uses the all-mpnet-base-v2 embedding model with 768 dimensions so that stored documents and memories can be represented as fixed-length vectors for retrieval.

### Vector Storage (Qdrant)

— PYRANA_PLATFORM_ARCHITECTURE_REFERENCE.md

- Embedding model: all-mpnet-base-v2 (768 dimensions)

— PYRANA_PLATFORM_ARCHITECTURE_REFERENCE.md

- Collections: docs_{region}, user_memory_{region}, session_memory_{region}

— PYRANA_PLATFORM_ARCHITECTURE_REFERENCE.md