Overview

SciPy 1.17, released on January 10, 2026, introduces N-dimensional batching and ILP64 BLAS support.

Main Features

N-dimensional batching

scipy.linalg functions now support N-dimensional batching, allowing batches of matrices to be processed in a single call.

python
import numpy as np
from scipy import linalg

# Batch of 100 3x3 matrices
matrices = np.random.randn(100, 3, 3)

# Batch inversion (N-dim batching)
inverses = linalg.inv(matrices)
print(inverses.shape)  # (100, 3, 3)

ILP64 BLAS

ILP64 BLAS support enables working with matrices whose dimensions exceed 2 billion elements.

Sources