#1212th of 17 in Vector databases
lance
Columnar table format for multimodal AI data with vector and full-text search
- Stars
- 7.3k
- License
- Apache-2.0
- Last commit
- Nov 2026
- Last release
- Oct 2026
- Language
- Rust
6,303 on 21 Apr 20267,279 on 10 Nov 2026
From GitHub's daily star counts.
Checked on GitHub.
Overview
Lance is a file format, table format and catalog spec written in Rust, with Python (pylance) and Java bindings. It stores images, video, audio, text and embeddings in one dataset on local disk or object storage, with IVF_PQ vector indices, BM25 full-text search and versioned tables with time travel. It integrates with Arrow, Pandas, Polars, DuckDB, Spark, Ray, Trino and Flink.
Who it is for: Engineers building ML data pipelines and vector search on object storage
Strengths
- Vector, BM25 full-text and SQL queries run on the same dataset with secondary indices
- Reads and writes S3, Azure, GCS, Hugging Face Hub and other stores by URI scheme
- Adds columns with backfilled values without rewriting the whole table
- Built-in ACID versioning with tags, branches and time travel
Weaknesses
- Library and format, not a server; the README shows no network query API
- Only stable data_storage_version values carry a compatibility guarantee; next is unstable
- Mixed Lance versions require pinning data_storage_version for deterministic writes
- Performance claims such as 100x vs Parquet come from the project's own benchmarks
What it needs
- Compose
Quick answers
Is lance open source?
Yes. Its code is on GitHub under the Apache-2.0 license.
Can you run lance with Docker?
Yes. Its repository has a Docker setup: a Dockerfile, a Compose file or a published image.
Also in Vector databases
See all 17| Rank | Project | Score |
|---|---|---|
| 1 | MilvusDistributed vector database with dense, sparse and hybrid search at scale | 90 out of 100 |
| 2 | MeilisearchRust search engine with typo tolerance, filtering and hybrid semantic search | 84 out of 100 |
| 3 | QdrantRust vector database with payload filtering, REST and gRPC | 84 out of 100 |
| #4 | faissC++ library for similarity search and clustering of dense vectors | 81 out of 100 |
| #5 | leannVector index that recomputes embeddings on demand to cut storage | 77 out of 100 |