Back to Gap Map Infrastructure / Storage & Retrieval

Lance

open source / Overall score: 4.0(strong)

Lance is a columnar table format for multimodal data that keeps random access fast enough to train and query directly against stored rows, rather than requiring a scan-oriented layout. Vector indexes, full-text indexes and dataset versioning are part of the format, and it interoperates with Pandas, DuckDB, Polars, PyArrow, Ray and Spark.

Openness

5 high confidence

5.0

license: Apache-2.0(OSI)
source: public
core features withheld: no

Apache-2.0 license body confirmed. The repository is public and unarchived and builds the whole product, and the README describes no paid tier, enterprise edition or license-gated build beside it, so source is public and the core ungated.

https://lance.org/ recorded 2026-09-30
Lance home page footer says "Lance is an open-source columnar format for multimodal AI. Apache-2.0."
https://pypi.org/pypi/pylance/json recorded 2026-09-30
PyPI metadata for pylance lists the classifier "License :: OSI Approved :: Apache Software License".
https://raw.githubusercontent.com/lance-format/lance/main/LICENSE recorded 2026-09-30
LICENSE file is the Apache-2.0 text: "Apache License" and "Version 2.0, January 2004" and "TERMS AND CONDITIONS FOR USE, REPRODUCTION, AND DISTRIBUTION".
https://raw.githubusercontent.com/lance-format/lance/main/README.md recorded 2026-09-30
README describes the format as "High-performance vector search, full-text search, random access, and feature engineering capabilities for the lakehouse." Searched the body for enterprise, paid tier, license key and cloud: none.

Adoption

4 high confidence

4.0

PyPI downloads of pylance drive this reading, since that package is the channel Lance declares for distribution. The figure therefore measures use of the format itself, not a proxy for it.

https://pypistats.org/api/packages/pylance/recent recorded 2026-08-18
last_month downloads = 4,157,433 for pylance

Capability

4 medium confidence

4.0

Lance is a columnar table format with its own vector and full-text indexes, zero-copy versioning and fast random access, readable by several analytical engines. It covers most of the storing-and-indexing job itself but ships no serving layer of its own, which keeps it one step below Vespa.

https://lance.org/ recorded 2026-09-30
Lance home page calls it "A file format, table format, and catalog spec for building a complete lakehouse on object storage" and lists Spark, Ray, PyTorch, Trino and DuckDB integrations.
https://raw.githubusercontent.com/lance-format/lance/main/README.md recorded 2026-09-30
README lists "Combine vector similarity search, full-text search (BM25), and SQL analytics on the same dataset with accelerated secondary indices." and "Automatic versioning with ACID transactions, time travel, tags, and branches".

Verified 2026-09-24