Deeplake

GPU-accelerated database for multimodal AI. Store, index, search, and stream data to your models and agents.

---

Get Started

• Python SDK Quickstart

  ---

  Install the SDK, create a table, ingest files, and run your first vector search in 2 minutes.

  Get started

• REST API Quickstart

  ---

  Call Deeplake from any language with plain HTTP. Create tables, upload files, and query with curl.

  Get started

Learn

• Guide

  ---

  Core concepts, tables, files, indexes, and search. Everything you need to build with Deeplake.

  Read the guide

• Examples

  ---

  Production-ready projects: image search, video retrieval, RAG, agent memory, and more.

  Browse examples

Reference

• REST API

  ---

  All HTTP endpoints: auth, workspaces, tables, SQL queries, files, and batch operations.

• SQL Operators

  ---

  Vector <#>, BM25, hybrid search, JSONB operators, index syntax, and tensor types.

• SDK Client

  ---

  Client constructor, ingest(), query(), open_table(), table management, and environment variables.

Open Source

• Core

  ---

  Self-hosted Deeplake: store datasets on S3, GCS, or Azure. Stream to PyTorch and TensorFlow. Version your data with Git-like commits.

  Core docs

---

Community · GitHub · LLM Docs