Oracle Simple Vector SDK - Python APIs

Oracle Simple Vector SDK brings together the APIs and tools for building vector search applications with Oracle AI Database. This guide covers its Python APIs, delivered by oracle-vecdb, the Oracle VecDB Python SDK.

Start here

Follow these two steps to get your first Python vector search result:

1. Set up your database. Choose a deployment, create a vector user, load an embedding model, and record the vector-user credentials, Console URL, and SDK REST endpoint.
2. Run the quickstart. Install oracle-vecdb, configure the Python client, create a table, load records, and run a semantic search.

The quickstart uses username and password authentication. For bearer-token configuration, see Configuration.

For a high-level explanation of the workflow and its terms, see Oracle AI Vector Search.

Continue with the Python SDK

After the quickstart or a sample, choose the task that matches your application:

• Configure the Python client, including authentication, TLS, and retry options.
• Create and load vector data, including integrated embeddings or bring-your-own vectors.
• Search results, including using metadata filters.

Explore example applications and notebooks

Explore the complete example applications and notebooks, including screenshots, sample descriptions, AI build prompts, and focused notebook workflows.

Learn the concepts

Use Concepts when you want to understand a design choice or see how the pieces of a vector search application fit together. Start with Embedding Workflows, Embedding and Reranking Models, or Oracle AI Vector Search.

Use other Oracle VecDB interfaces

This guide focuses on the Python SDK. For REST APIs, PL/SQL APIs, and the Vector Database Console, see Access Oracle AI Vector Search.

Find exact API details

Use the Python API reference for exact signatures, parameters, return values, response objects, and errors. For connection and service problems, see Troubleshooting.