0.2.7
Released on 2026-05-05.
This is a re-release of all mkl-service 2.4.x, 2.5.x, and 2.6.x wheels, fixing a missing upper bound on the mkl runtime dependency. The unpatched wheels can be downloaded from the original release pages.
Latest builds
| Package | OS | Latest version | Supported Python versions |
|---|---|---|---|
| NumPy | Linux | 2.4.4 | ≥3.11, ≤3.14 |
| NumPy | Windows | 2.4.4 | ≥3.11, ≤3.14 |
| SciPy | Linux | 1.17.1 | ≥3.11, ≤3.14 |
| SciPy | Windows | 1.17.1 | ≥3.11, ≤3.14 |
| mkl-service | Linux | 2.7.0 | ≥3.11, ≤3.14 |
| mkl-service | Windows | 2.7.0 | ≥3.11, ≤3.14 |
Usage
It is recommended to use uv or pip to automatically handle all dependencies.
uv
# Run this from project directory
uv init
uv add numpy scipy --index https://urob.github.io/numpy-mklpip
pip install numpy scipy --index-url https://urob.github.io/numpy-mklManual installation
Current wheels for all packages and versions can be downloaded here. Attached below are newly built wheels only.
Manual installs must install compatible versions of mkl, mkl-service, and their indirect dependencies. For maximum compatibility, it is recommended to install mkl-service from this repository, which has been patched to automatically detect and load the mkl library whenever numpy or scipy are imported.