# google/tensorstore

Library for reading and writing large multi-dimensional arrays.

Repository: https://github.com/google/tensorstore
Canonical: https://ross.abutalabs.com/products/tensorstore
Homepage: https://google.github.io/tensorstore/
Language: C++
License: NOASSERTION
License Family: other
Last push: 2026-08-25T01:08:44+00:00

## Health v2 (maintenance only)
Score: 77/100 (v2, computed 2026-09-03T02:20:16.233290+00:00)
- activity 99, release rhythm 35, longevity 100
- inputs: {"age_days": 2347, "days_push": 9, "days_rel": null, "gap_med": null, "n_releases_24m": 0}
- flags: no_releases, no_license
- formula: round(0.45*activity + 0.35*rhythm + 0.20*longevity); archived -> min(score, 10)

## Adoption (not part of the score)
Stars 1536, forks 144 (observed 2026-08-28T04:04:59.905890+00:00)

## What it is
TensorStore is a C++ and Python library for reading and writing large multi-dimensional arrays with a uniform API across formats like zarr, N5, and Neuroglancer precomputed. It supports multiple storage backends (local filesystems, Google Cloud Storage, S3), asynchronous high-throughput access, caching, transactions with ACID guarantees, and safe concurrent access from multiple processes and machines.

## Use cases
- store and read large multi-dimensional arrays like image volumes or ML datasets
- read and write zarr or N5 arrays from Python
- access chunked array data on S3 or Google Cloud Storage
- process large arrays concurrently from multiple machines safely
- stream large arrays with high throughput over high-latency remote storage
- apply composable indexing and virtual views to array data

## When to choose
- you need to store or manipulate very large n-dimensional arrays that exceed memory
- you need a uniform API over multiple array storage formats and backends
- you need transactional, consistent concurrent access to array data across processes or machines
- you work in scientific computing, neuroimaging, or ML data pipelines with chunked array storage

## When to avoid
- you only need small in-memory arrays handled by NumPy or PyTorch tensors
- you need a general-purpose relational or document database
- you need tabular or dataframe storage rather than multi-dimensional arrays
- you need a simple key-value store without array semantics

## Facets
- artifact type: library
- maturity: active
- function: database-driver, serialization, file-system, caching, concurrency
- domain: data-science, machine-learning, big-data, developer-tools
- platform: cpp, python, windows, cross-platform
- tags: multi-dimensional-arrays, zarr, n5, tensor-storage, object-storage, acid-transactions, scientific-computing, data-engineering, linux, macos

## Member repositories
- google/tensorstore (main) score 77

## Provenance
- Observed fields: from GitHub, fetched 2026-08-28T04:04:59.905890+00:00.
- Health v2: computed from the inputs above; adoption is never an input.
- Inferred fields (summary, facets, guidance): AI-extracted, prompt v1, taxonomy v1, on 2026-08-30T04:31:04.658075+00:00, confidence not recorded.
  - readme: https://github.com/google/tensorstore (fetched 2026-08-28T04:04:59.905890+00:00, sha 508f453d6b0e)
  - homepage: https://google.github.io/tensorstore/ (fetched 2026-08-29T11:32:47.131244+00:00, sha 0294038131a2)
- Data as of 2026-08-30T08:39:29.467469+00:00.