Freedom251 › Data Version Control Tools
Lance
What Lance is
Lance is an open source lakehouse format for multimodal AI data stored on object storage. Its file, table, and catalog specifications bring images, video, audio, text, embeddings, and tabular data into one format. Search on a dataset can combine vector similarity, BM25 full-text search, and SQL analytics, accelerated by secondary indexes. The project says random access is 100× faster than Parquet and Iceberg. The table format supports ACID transactions, schema evolution, time travel, dataset versioning, and incremental updates through MVCC. Adding columns with backfill can use new Lance files instead of rewriting the full table. Lance can commit directly to object storage; its Directory Catalog supports deployment there without extra infrastructure. Listed integrations include Pandas, Polars, Ray, PyTorch, DataFusion, DuckDB, Spark, Trino, Flink, Apache Polaris, Unity Catalog, Apache Gravitino, and Hive Metastore. Lance is self-hosted and free under Apache-2.0. Its Python SDK is pylance, installed with `pip install pylance`. The current unstable Lance 2.3 file format has no compatibility guarantee and the documentation says not to use it in production.
Who it suits
Lance suits teams working with multimodal datasets on object storage that need search, SQL analytics, or versioned table updates. Python users can start with the pylance SDK; production users should note the stated restriction on the unstable 2.3 format.
What it does well
- Stores multimodal and tabular data in one format.
- Combines vector, full-text, and SQL search.
- Supports transactions, schema evolution, and time travel.
- Can commit directly to object storage.
- Free and open source under Apache-2.0.
What to know first
- Self-hosted deployment model.
- Unstable Lance 2.3 has no compatibility guarantee.
- Documentation says not to use unstable 2.3 in production.
Plans
| Plan | Billed and limits | Price |
|---|---|---|
| Lance | Open source · Apache-2.0 | Free |
As lance.org lists them, read 8 Oct 2026. The maker's own page has the final word.
Verdict
Lance brings multimodal data, search, and table management together for object-storage workflows. Its stated unstable 2.3 format is not recommended for production, so check which format version is in use.
Facts from Lance's own pages
- Catalogs
- Catalog implementations listed include Apache Hive, Apache Polaris, Apache Gravitino, Unity Catalog, and AWS Glue. · lance.org, 8 Oct 2026
- Community support
- The project links users to Discord and a mailing list for community communication. · lance.org, 8 Oct 2026
- Data management
- The table format supports ACID transactions, schema evolution, time travel, and incremental updates through MVCC. · lance.org, 8 Oct 2026
- Getting started
- The quickstart identifies the Python SDK, installed with `pip install pylance`, as the easiest way to start using Lance. · lance.org, 8 Oct 2026
- Incubating integrations
- The governance page identifies Lance Context, Data Viewer, Flink, Graph, Trino, and pglance as incubating projects whose APIs or functionality may change. · lance.org, 8 Oct 2026
- Integrations
- Documented integrations include DataFusion, Spark, DuckDB, Trino, PyTorch, TensorFlow, Ray, Hugging Face, PostgreSQL, and Flink. · lance.org, 8 Oct 2026
- License
- The site identifies Lance as open source under the Apache-2.0 license. · lance.org, 8 Oct 2026
- Multimodal data
- Lance stores images, videos, audio, text, and embeddings alongside tabular data in one format. · lance.org, 8 Oct 2026
- Notable limit
- The current unstable Lance 2.3 file format has no compatibility guarantee and the documentation says it should not be used in production. · lance.org, 8 Oct 2026
- Product
- Lance is an open source lakehouse format for multimodal AI, with file, table, and catalog specifications for object storage. · lance.org, 8 Oct 2026
- Purpose
- Lance is an open lakehouse format for multimodal AI, built for data on object storage. · lance.org, 8 Oct 2026
- Python SDK
- The quickstart identifies pylance as the Python SDK and gives `pip install pylance` as the installation command. · lance.org, 8 Oct 2026
- Random access
- The site says Lance offers 100× faster random access than Parquet and Iceberg. · lance.org, 8 Oct 2026
- Schema evolution
- Lance supports adding columns with backfill by writing new Lance files rather than rewriting the full table. · lance.org, 8 Oct 2026
- Search
- Lance combines vector similarity, BM25 full-text search, and SQL analytics on the same dataset, accelerated by secondary indexes. · lance.org, 8 Oct 2026
- Security reporting
- The community page asks users to report security vulnerabilities privately to the PMC mailing list and avoid public disclosure until the issue is resolved. · lance.org, 8 Oct 2026
- SQL access
- Lance datasets can be queried with SQL through integrations including DataFusion, Spark, DuckDB, and Trino. · lance.org, 8 Oct 2026
- Storage
- Lance can commit directly to object storage, and its Directory Catalog supports deployments directly on object stores without additional infrastructure. · lance.org, 8 Oct 2026
- Support
- The contributor guide directs users with questions to the Lance Discord for real-time support. · lance.org, 8 Oct 2026
- Versioning
- The format supports ACID transactions, time travel, and dataset versioning. · lance.org, 8 Oct 2026
Where Lance stands on Freedom251
- Data Version Control Tools#30 of 36