Infrastructure / Storage & Retrieval

SPTAG

SPTAG (Space Partition Tree And Graph) is a Microsoft Research and Bing library for large-scale approximate nearest neighbour search, combining kd-tree or balanced k-means tree partitioning with a relative neighbourhood graph. It supports online insertion and deletion and a distributed search mode across machines.

Distinct from diskann - a different Microsoft Research ANN algorithm family (tree-plus-graph rather than Vamana), maintained independently. sptag on PyPI is declared as the measurement identity because, unlike diskannpy, it is built from this repository's root `setup.py` against the current code by the Microsoft SPTAG team, and the library has no server for it to be a client of; installing it gives you SPTAG. Its count is small because most users build the C++ library directly, which the adoption note records. Verified 2026-09-02 via the GitHub API, the PyPI metadata (Homepage points back at Microsoft/SPTAG) and the LICENSE body.

Openness

5 high confidence

5.0

license: MIT(OSI)
source: public
core-gated: ungated

MIT license body confirmed. The repository is public and unarchived and builds the whole library, and the README describes no paid tier, enterprise edition or license-gated build beside it.

https://github.com/microsoft/SPTAG/blob/main/LICENSE recorded 2026-09-02
LICENSE file is the verbatim MIT text under a Microsoft Corporation copyright line.
https://api.github.com/repos/microsoft/SPTAG recorded 2026-09-02
Repo metadata - license spdx_id MIT, private false, archived false, default branch main - for microsoft/SPTAG.
https://github.com/microsoft/SPTAG/blob/main/README.md recorded 2026-09-02
README describes an approximate nearest neighbour search library with no paid tier, enterprise edition or license-key-gated build beside the MIT-licensed code.

Adoption

1 medium confidence

1.0

77 PyPI downloads of `sptag` in the trailing 30 days, which lands in the <10K band of the software usage scale, level 1. sptag is a niche C++ library with Python bindings and no separate server, so the package figure is a direct if small read of usage.

https://pypistats.org/api/packages/sptag/recent recorded 2026-09-02
last_month downloads = 77 for sptag.

Capability

3 medium confidence

3.0

Banded on the category feature matrix as working single-purpose store or index, level with faiss - more than a single index family shipped as a bare library (it offers two partitioning strategies, online updates and a distributed serving mode), but it is still a search toolkit rather than a durable, transactional store, so it does not reach the complete-system band above.

https://github.com/microsoft/SPTAG/blob/main/README.md recorded 2026-09-02
README documents two index families, online insert/delete, and a distributed serving mode across multiple machines.

Verified 2026-09-02