#1414th of 15 in Vector databases
deeplake
Serverless database for vectors and raw AI data, with versioning and ML dataloaders
Documentation ↗
(opens in a new tab)Website ↗
(opens in a new tab)Repository on GitHub ↗
(opens in a new tab)
- Stars
- 9.3k
- License
- Apache-2.0
- Last commit
- May 2026
- Last release
- Feb 2026
- Language
- C++
9,007 on 1 Apr 20269,258 on 21 Oct 2026
From GitHub's daily star counts.
Checked on GitHub.
Overview
Deep Lake is a Python-installed database that stores embeddings alongside raw data such as images, video, audio, text and PDFs in a chunked columnar format. It serves as a vector store for LLM apps (LangChain, LlamaIndex) and as a streaming dataset layer for PyTorch and TensorFlow training. Computation runs client-side, and data can live on S3, GCP, Azure, S3-compatible storage like MinIO, or local disk.
Who it is for: ML engineers storing multimodal datasets and embeddings for RAG or training
Strengths
- Stores raw media and embeddings together, not just vectors with light metadata
- Built-in dataset version control and PyTorch/TensorFlow dataloaders
- Serverless and client-side; works with S3, GCP, Azure, MinIO, local or in-memory storage
- Integrates with LangChain, LlamaIndex, Weights & Biases and MMDetection
Weaknesses
- README says full features require registering in the hosted Deep Lake App
- Visualizer and public dataset hub are tied to Activeloop's cloud service
- No server mode; README gives no benchmarks for large-scale vector search
- README links mix old and new docs sites, and comparisons look dated
What it needs
- no GPU
- Needs Python (pip install deeplake), Deep Lake App account for full features, Optional S3/GCP/Azure storage
- Models: PyTorch, TensorFlow, LangChain, LlamaIndex
Also in Vector databases
See all 15| Rank | Project | Score |
|---|---|---|
| 1 | MilvusDistributed vector database with dense, sparse and hybrid search at scale | 83 out of 100 |
| 2 | MeilisearchRust search engine with typo tolerance, filtering and hybrid semantic search | 75 out of 100 |
| 3 | QdrantRust vector database with payload filtering, REST and gRPC | 73 out of 100 |
| #4 | lancedbEmbedded vector database built on the Lance columnar format | 69 out of 100 |
| #5 | faissC++ library for similarity search and clustering of dense vectors | 67 out of 100 |