S3-compatible · Binary-level deduplication · Kubernetes-native · Apache 2.0
UltiHash Core is an open-source object storage system that eliminates redundant data at the binary level — across all objects, all formats, continuously. It speaks S3, deploys on Kubernetes via Helm, and reduces storage requirements by up to 60% on typical AI datasets without any changes to your application.
| Repository | What it is |
|---|---|
| core | The storage engine — C++, CMake, Apache 2.0 |
| uh-helm | Helm chart for Kubernetes deployment |
| uh-documentation | Source for docs.ultihash.io |
| scripts | Integration examples — PySpark, Trino, Airflow, boto3 and more |
| core-tests | Integration test suite |
Try it in 60 seconds — no account needed:
curl -fsSL https://ultihash.io/1line | bashSpins up a local Docker cluster, lets you point it at any directory, and shows you the deduplication savings in your terminal.
Made in Berlin · ultihash.io · Join us on Discord
