LanceDB

By LanceDB

Embedded vector database built on the Lance columnar format — serverless and local-first

Website ↗ Repository ↗ Docs ↗

python-v0.35.0-beta.22026-07-14Rust / PythonApache-2.0

#vector-db #embeddings #embedded #rust

11.0k Open source Free tier Paid

Doc version 1.0 · Updated June 28, 2026

Related tools

Overview

LanceDB is an open-source, embedded vector database built on the Lance columnar format — it runs in-process and writes to local files or object storage, with no server to operate.
It is multimodal and fast, and scales from a laptop to LanceDB Cloud without changing the API.

flowchart LR
  rows[rows + vectors] --> t[(LanceDB table)]
  q[query vector] --> t
  t --> top[nearest rows]

The Code samples tab shows the fully embedded flow.

When to use it

Choose LanceDB when you want a zero-ops, local-first vector store embedded in your app — ideal for desktop tools, notebooks, and edge deployments, with a cloud option when you outgrow local files.

Pricing

Open source (Apache-2.0), free to self-host. LanceDB Cloud is usage-based.

Plan	Price	Notes
Cloud (free)	$0	Usage-based free tier
Enterprise	Custom	Annual, dedicated

Source ↗ · as of 2026-06-27

Example

Python · embedded

When to use

Reach for LanceDB when you want an embedded vector store with no server — it writes to disk like SQLite, but for vectors.

How it flows

Connect to a local directory, create a table of rows with vectors, then search by a query vector for the nearest rows.

Structure

flowchart LR
  rows[rows + vectors] --> t[(LanceDB table)]
  q[query vector] --> t
  t --> top[nearest rows]

import lancedb

db = lancedb.connect("./lance")   # embedded; data lives in a local directory
tbl = db.create_table("docs", data=[
    {"vector": [0.1, 0.2, 0.3, 0.4], "text": "hello"},
    {"vector": [0.9, 0.8, 0.7, 0.6], "text": "world"},
])

hits = tbl.search([0.1, 0.2, 0.3, 0.4]).limit(1).to_list()
print(hits[0]["text"])

Example

LanceDB — embedded vector search

A real, runnable mini-project. Download it and run with Docker.

View samples/lancedb_1 on GitHub ↗

June 28, 2026

LanceDB — embedded vector search

A tiny LanceDB script: it creates a table of vectors, runs a nearest-neighbour search, and prints the closest row. Fully embedded — no server, no API key, no network — so the output is deterministic.

Run with Docker

cd samples/lancedb_1
docker build -t aas-lancedb .
docker run --rm aas-lancedb

Run with Docker (in a devcontainer with DooD)

In a dev container that talks to the host Docker daemon (Docker-outside-of-Docker), the foreground docker run above often prints nothing and exits 0 — but the run itself succeeds. Run detached and follow the logs instead:

cd samples/lancedb_1
docker build -t aas-lancedb .
docker logs -f "$(docker run -d aas-lancedb)"

Run locally

cd samples/lancedb_1
pip install -r requirements.txt
python app.py

Example run

LanceDB runs locally and this sample is deterministic, so the output is the same every run.

nearest: Cats are independent pets that purr.

Files

app.py

"""LanceDB: an embedded vector database — no server, data on local disk.

Creates a small table of vectors, runs a nearest-neighbour search, and prints
the closest row. Fully local and deterministic: no API key, no network.

Run:
    docker build -t aas-lancedb .
    docker run --rm aas-lancedb
"""

import lancedb


def main() -> None:
    db = lancedb.connect("/tmp/lance-demo")  # embedded; a local directory
    db.drop_table("docs", ignore_missing=True)
    tbl = db.create_table(
        "docs",
        data=[
            {"vector": [0.1, 0.2, 0.3, 0.4], "text": "Cats are independent pets that purr."},
            {"vector": [0.9, 0.8, 0.7, 0.6], "text": "A rocket is propelled by ejecting exhaust."},
        ],
    )

    hits = tbl.search([0.12, 0.21, 0.29, 0.41]).limit(1).to_list()
    print("nearest:", hits[0]["text"])


if __name__ == "__main__":
    main()

Dockerfile

# Minimal image for the LanceDB sample. Fully embedded — no services, no keys.
FROM python:3.13-slim

ENV PYTHONUNBUFFERED=1

WORKDIR /app

COPY requirements.txt .
RUN pip install --no-cache-dir -r requirements.txt

COPY app.py .

ENTRYPOINT ["python", "app.py"]