Hypre is a software library for solving large, sparse linear systems on massively parallel computers, focusing on modern, scalable preconditioners. It provides conceptual interfaces to allow users to access the library in a way that aligns with their problem-solving approach. The library includes various preconditioners and solvers, such as conjugate gradient, GMRES, and multigrid methods, and supports different interfaces for structured, semi-structured, finite element, and linear-algebraic problems. The library's conceptual interfaces simplify the process of accessing solvers by abstracting away the need for complex sparse matrix structures. Users define the problem in terms of grids, stencils, or finite elements, and the library handles the underlying data structures. This approach reduces coding burden and allows for more efficient and scalable solutions. Hypre supports four conceptual interfaces: structured-grid, semi-structured-grid, finite-element, and linear-algebraic. These interfaces cater to different types of problems and allow users to access a wide range of powerful preconditioners. For example, geometric multigrid (GMG) requires structured grids, while algebraic multigrid (AMG) uses finite element information. The library includes several preconditioners, including SMG, PFMG, BoomerAMG, ParaSails, PILUT, and Euclid. These preconditioners are designed to be efficient and scalable, with BoomerAMG being a parallel algebraic multigrid method that can handle both structured and unstructured grids. ParaSails is a parallel sparse approximate inverse preconditioner, and Euclid is a scalable implementation of the Parallel ILU algorithm. Hypre is designed to be flexible and interoperable, allowing it to be used as a solver package or as a framework for algorithm development. It is written in C but can also be used from Fortran. The library is available for download and includes detailed documentation and user manuals. Future work includes adding new preconditioners and improving existing ones, such as AMGe, an algebraic multigrid method based on local finite element stiffness matrices. Research is also ongoing to enhance BoomerAMG with multi-coloring techniques and other methods to improve parallel performance and convergence.Hypre is a software library for solving large, sparse linear systems on massively parallel computers, focusing on modern, scalable preconditioners. It provides conceptual interfaces to allow users to access the library in a way that aligns with their problem-solving approach. The library includes various preconditioners and solvers, such as conjugate gradient, GMRES, and multigrid methods, and supports different interfaces for structured, semi-structured, finite element, and linear-algebraic problems. The library's conceptual interfaces simplify the process of accessing solvers by abstracting away the need for complex sparse matrix structures. Users define the problem in terms of grids, stencils, or finite elements, and the library handles the underlying data structures. This approach reduces coding burden and allows for more efficient and scalable solutions. Hypre supports four conceptual interfaces: structured-grid, semi-structured-grid, finite-element, and linear-algebraic. These interfaces cater to different types of problems and allow users to access a wide range of powerful preconditioners. For example, geometric multigrid (GMG) requires structured grids, while algebraic multigrid (AMG) uses finite element information. The library includes several preconditioners, including SMG, PFMG, BoomerAMG, ParaSails, PILUT, and Euclid. These preconditioners are designed to be efficient and scalable, with BoomerAMG being a parallel algebraic multigrid method that can handle both structured and unstructured grids. ParaSails is a parallel sparse approximate inverse preconditioner, and Euclid is a scalable implementation of the Parallel ILU algorithm. Hypre is designed to be flexible and interoperable, allowing it to be used as a solver package or as a framework for algorithm development. It is written in C but can also be used from Fortran. The library is available for download and includes detailed documentation and user manuals. Future work includes adding new preconditioners and improving existing ones, such as AMGe, an algebraic multigrid method based on local finite element stiffness matrices. Research is also ongoing to enhance BoomerAMG with multi-coloring techniques and other methods to improve parallel performance and convergence.