LanceDB

aid: lancedb
url: https://raw.githubusercontent.com/api-evangelist/lancedb/refs/heads/main/apis.yml
name: LanceDB
kind: opensource
description: >-
  LanceDB is the AI-Native multimodal lakehouse built on the open-source Lance
  columnar storage format. It pairs an Apache 2.0 licensed embedded retrieval
  library (Python, TypeScript, Rust, Go, C, Java SDKs) with a managed cloud
  service (LanceDB Cloud) and an enterprise lakehouse (LanceDB Enterprise) that
  unify vector, full-text, hybrid, and SQL search across billions of multimodal
  records. The REST surface is governed by the open Lance Namespace
  specification (OpenAPI 3.1) covering namespace, table, index, tag, and
  transaction operations with first-class support for materialized views,
  schema evolution, and time-travel versioning. LanceDB is used in production
  by Midjourney, Runway, World Labs, Netflix, Character.AI, Uber, NVIDIA,
  ByteDance, Databricks, and others for RAG, agent memory, training data
  curation, feature engineering, and large-scale retrieval.
image: https://kinlane-productions2.s3.amazonaws.com/api-evangelist-network/lancedb/lancedb-logo.png
tags:
  - Vector Database
  - Multimodal
  - Lance Format
  - Lakehouse
  - RAG
  - Agent Memory
  - Open Source
  - Embeddings
  - Full-Text Search
  - Hybrid Search
  - Columnar Storage
  - Arrow
  - AI Infrastructure
created: '2026-05-23'
modified: '2026-05-25'
specificationVersion: '0.20'
apis:
  - aid: lancedb:oss
    name: LanceDB OSS
    description: >-
      Apache 2.0 licensed, developer-friendly, embedded multimodal retrieval
      library. Provides namespace + table CRUD, insert / upsert / delete,
      vector / full-text / hybrid / SQL search, scalar (BTree, Bitmap,
      LabelList, FTS) and vector (IVF_FLAT, IVF_PQ, IVF_HNSW_SQ) indexes,
      automatic versioning, time-travel, and pluggable embedding functions.
      Backed by S3, GCS, Azure Blob, or local disk via object_store.
    humanURL: https://docs.lancedb.com/
    baseURL: https://github.com/lancedb/lancedb
    tags:
      - Open Source
      - Embedded
      - Vector Database
      - Apache 2.0
    properties:
      - type: Documentation
        url: https://docs.lancedb.com/
      - type: Repository
        url: https://github.com/lancedb/lancedb
      - type: Quickstart
        url: https://docs.lancedb.com/quickstart
      - type: License
        url: https://github.com/lancedb/lancedb/blob/main/LICENSE
      - type: Releases
        url: https://github.com/lancedb/lancedb/releases
      - type: Issues
        url: https://github.com/lancedb/lancedb/issues
      - type: OpenAPI
        url: https://raw.githubusercontent.com/api-evangelist/lancedb/main/openapi/lance-namespace-openapi.yaml
  - aid: lancedb:cloud
    name: LanceDB Cloud
    description: >-
      Fully managed serverless LanceDB service. Hosts customer namespaces and
      tables, exposes the Lance Namespace REST API, and handles compaction,
      index maintenance, replication, and elasticity. Authenticated via
      `x-api-key` header. SOC 2, GDPR, and HIPAA compliant.
    humanURL: https://lancedb.com/
    baseURL: https://your-tenant.lancedb.cloud/v1/
    tags:
      - Managed Service
      - Serverless
      - REST
      - Multi-Tenant
    properties:
      - type: Documentation
        url: https://docs.lancedb.com/
      - type: API
        url: https://docs.lancedb.com/api-reference/rest/
      - type: OpenAPI
        url: https://raw.githubusercontent.com/api-evangelist/lancedb/main/openapi/lance-namespace-openapi.yaml
      - type: Trust
        url: https://trust.lancedb.com/
      - type: Sales
        url: https://lancedb.com/contact
  - aid: lancedb:enterprise
    name: LanceDB Enterprise
    description: >-
      Distributed, multi-tenant multimodal lakehouse. Adds curation and
      deduplication, Python UDF feature engineering, materialized views,
      GPU-accelerated index build via cuVS, distributed query, and direct
      training integration with PyTorch / Ray. Deployable in customer VPC on
      AWS, GCP, or Azure.
    humanURL: https://lancedb.com/
    baseURL: https://lancedb.com/
    tags:
      - Enterprise
      - Managed
      - Lakehouse
      - VPC
      - GPU
    properties:
      - type: Documentation
        url: https://docs.lancedb.com/enterprise/
      - type: Architecture
        url: https://docs.lancedb.com/enterprise/architecture
      - type: Sales
        url: https://lancedb.com/contact
      - type: Trust
        url: https://trust.lancedb.com/
  - aid: lancedb:lance-format
    name: Lance Format
    description: >-
      Apache 2.0 open lakehouse format for multimodal AI. Columnar Parquet
      replacement offering 100x faster random access, zero-copy reads, vector
      indexes, and automatic data versioning. Convertible from Parquet in two
      lines of code. Maintained by the lance-format community.
    humanURL: https://lance.org/
    baseURL: https://github.com/lance-format/lance
    tags:
      - File Format
      - Columnar
      - Arrow
      - Open Source
      - Apache 2.0
    properties:
      - type: Documentation
        url: https://lance.org/
      - type: Repository
        url: https://github.com/lance-format/lance
      - type: Specification
        url: https://lance.org/format/
      - type: Releases
        url: https://github.com/lance-format/lance/releases
  - aid: lancedb:lance-namespace
    name: Lance Namespace Specification
    description: >-
      Open OpenAPI 3.1 specification describing access and operations against
      a collection of Lance tables in a multimodal lakehouse. Defines a
      complete REST server contract (50+ operations) covering namespaces,
      tables, indices, tags, transactions, materialized views, schema
      evolution, and time-travel. Reference implementations exist for Apache
      Hive, Apache Polaris, Apache Gravitino, Unity Catalog, and AWS Glue.
    humanURL: https://lance.org/format/namespace/
    baseURL: https://github.com/lance-format/lance-namespace
    tags:
      - Specification
      - OpenAPI
      - REST
      - Catalog
      - Open Source
    properties:
      - type: Specification
        url: https://lance.org/format/namespace/
      - type: Repository
        url: https://github.com/lance-format/lance-namespace
      - type: OpenAPI
        url: https://raw.githubusercontent.com/lance-format/lance-namespace/main/docs/src/spec.yaml
      - type: Implementations
        url: https://github.com/lance-format/lance-namespace-impls
  - aid: lancedb:python-sdk
    name: LanceDB Python SDK
    description: >-
      Primary client library. First-class Arrow, Pandas, Polars, and Pydantic
      integration; pluggable embedding functions covering OpenAI, Cohere,
      Jina, Hugging Face, Ollama, Bedrock, Sentence Transformers and 20+ more.
    humanURL: https://lancedb.github.io/lancedb/python/python/
    baseURL: https://pypi.org/project/lancedb/
    tags:
      - SDK
      - Python
    properties:
      - type: Documentation
        url: https://lancedb.github.io/lancedb/python/python/
      - type: Package
        url: https://pypi.org/project/lancedb/
      - type: Repository
        url: https://github.com/lancedb/lancedb
  - aid: lancedb:typescript-sdk
    name: LanceDB TypeScript SDK
    description: >-
      Node.js / TypeScript / JavaScript client library for LanceDB OSS, Cloud,
      and Enterprise. Bundles native bindings via napi-rs.
    humanURL: https://lancedb.github.io/lancedb/js/globals/
    baseURL: https://www.npmjs.com/package/@lancedb/lancedb
    tags:
      - SDK
      - TypeScript
      - JavaScript
      - Node.js
    properties:
      - type: Documentation
        url: https://lancedb.github.io/lancedb/js/globals/
      - type: Package
        url: https://www.npmjs.com/package/@lancedb/lancedb
      - type: Repository
        url: https://github.com/lancedb/lancedb
  - aid: lancedb:rust-sdk
    name: LanceDB Rust SDK
    description: >-
      Native Rust client library; the LanceDB core and storage layer are
      written in Rust.
    humanURL: https://docs.rs/lancedb/latest/lancedb/
    baseURL: https://crates.io/crates/lancedb
    tags:
      - SDK
      - Rust
    properties:
      - type: Documentation
        url: https://docs.rs/lancedb/latest/lancedb/
      - type: Package
        url: https://crates.io/crates/lancedb
      - type: Repository
        url: https://github.com/lancedb/lancedb
  - aid: lancedb:go-sdk
    name: LanceDB Go SDK
    description: >-
      Official Go client library for LanceDB OSS, Cloud, and Enterprise.
    humanURL: https://github.com/lancedb/lancedb-go
    baseURL: https://github.com/lancedb/lancedb-go
    tags:
      - SDK
      - Go
    properties:
      - type: Repository
        url: https://github.com/lancedb/lancedb-go
  - aid: lancedb:c-sdk
    name: LanceDB C Bindings
    description: >-
      C ABI bindings for LanceDB enabling embedding into C, C++, and other
      FFI-capable hosts.
    humanURL: https://github.com/lancedb/lancedb-c
    baseURL: https://github.com/lancedb/lancedb-c
    tags:
      - SDK
      - C
      - FFI
    properties:
      - type: Repository
        url: https://github.com/lancedb/lancedb-c
  - aid: lancedb:mcp-server
    name: LanceDB MCP Server
    description: >-
      Model Context Protocol (MCP) server exposing LanceDB tables as
      retrieval tools for MCP-aware agents and IDEs.
    humanURL: https://github.com/lancedb/lancedb-mcp-server
    baseURL: https://github.com/lancedb/lancedb-mcp-server
    tags:
      - MCP
      - Agent
      - Tools
    properties:
      - type: Repository
        url: https://github.com/lancedb/lancedb-mcp-server
common:
  - type: Website
    url: https://lancedb.com/
  - type: Documentation
    url: https://docs.lancedb.com/
  - type: Blog
    url: https://lancedb.com/blog
  - type: GitHub
    url: https://github.com/lancedb
  - type: GitHubFormat
    url: https://github.com/lance-format
  - type: Discord
    url: https://discord.com/invite/G5DcmnZWKB
  - type: Trust
    url: https://trust.lancedb.com/
  - type: Support
    url: mailto:support@lancedb.com
  - type: Contact
    url: https://lancedb.com/contact
  - type: LLMsTxt
    url: https://docs.lancedb.com/llms.txt
  - type: OpenAPI
    url: https://raw.githubusercontent.com/api-evangelist/lancedb/main/openapi/lance-namespace-openapi.yaml
  - type: Capabilities
    url: https://raw.githubusercontent.com/api-evangelist/lancedb/main/capabilities/multimodal-retrieval.yaml
  - type: Vocabulary
    url: https://raw.githubusercontent.com/api-evangelist/lancedb/main/vocabulary/lancedb-vocabulary.yml
integrations:
  - name: Apache Arrow
  - name: Apache Iceberg
  - name: Apache Spark
  - name: Apache Flink
  - name: Apache Hive
  - name: Apache Polaris
  - name: Apache Gravitino
  - name: DuckDB
  - name: Pandas
  - name: Polars
  - name: Pydantic
  - name: Ray
  - name: Trino
  - name: PostgreSQL
  - name: Unity Catalog
  - name: AWS Glue
  - name: Hugging Face
  - name: LangChain
  - name: LlamaIndex
  - name: GenKit
  - name: PyTorch
  - name: Databricks
  - name: Amazon Web Services
  - name: Google Cloud
  - name: Microsoft Azure
  - name: NVIDIA
  - name: OpenAI
  - name: Cohere
  - name: Anthropic
  - name: Jina
  - name: Ollama
  - name: Sentence Transformers
  - name: CrewAI
  - name: cuVS
solutions:
  - name: Retrieval-Augmented Generation (RAG)
  - name: Long-Term Agent Memory
  - name: Training Data Curation
  - name: Feature Engineering for ML
  - name: Multimodal Search (Text, Image, Video, Audio)
  - name: Semantic Code Search
  - name: Recommendation Systems
  - name: Anomaly Detection
  - name: Vector Lakehouse
useCases:
  - name: Production RAG over multimodal corpora
  - name: Agent and assistant long-term memory store
  - name: Foundation-model training data curation and deduplication
  - name: Semantic search across billions of documents and embeddings
  - name: Hybrid (vector + BM25 + SQL) search with reranking
  - name: Time-travel and reproducible ML datasets via Lance versioning
  - name: GPU-accelerated vector index construction
maintainers:
  - FN: Kin Lane
    email: kin@apievangelist.com