MOSEK ApS is the company that offers an optimization suite that has been solving large-scale linear, quadratic, conic, and mixed-integer problems since the late 1990s.

MOSEK ApS was incorporated on 15 October 1997 in Copenhagen, and the first MOSEK solver was released in 1999. The company was founded by Erling Dalgaard Andersen and Knud D. Andersen.

Today, the MOSEK Optimization Suite handles linear, convex quadratic, conic quadratic (SOCP), semidefinite, exponential, and power-cone models as well as mixed-integer variants, and is prized for high-performance interior-point and mixed-integer solvers. Support for the nonsymmetric cones, i.e., the exponential and power cones, was introduced in recent versions to broaden conic modeling power.

MOSEK ships the Fusion object-oriented modeling API with a mature Python binding, provides low-level optimizer APIs and language interfaces for C, C#, Java, MATLAB, Python, R, Julia, and Rust, and plugs into modeling ecosystems and interfaces such as AMPL Optimization, GAMS, and CVXPY. The official docs and Fusion distribution include many worked examples and a modeling cookbook to help port and build models.

MOSEK is a commercial product with floating and server licensing but offers free academic licenses for students and faculty, plus a 30-day trial for evaluation. MOSEK has financial institutions, engineering firms, and software vendors among its users and publishes extensive docs, examples, and release notes to support adoption.