NVIDIA/cuopt

GPU accelerated decision optimization

What it solves

cuOpt is a GPU-accelerated optimization engine designed to solve complex mathematical programming and routing problems at scale. It addresses the challenge of finding high-quality solutions for large-scale linear and quadratic programming problems—some with millions of variables and constraints—in near real-time, which is often computationally prohibitive for traditional CPU-based solvers.

How it works

Written in C++ and exposed via C, Python, and Server APIs, cuOpt leverages NVIDIA GPUs to accelerate the solving process. It supports various optimization types, including Linear Programming (LP), Quadratic Programming (QP), and Vehicle Routing Problems (VRP). For Mixed Integer Programming (MIP), it uses GPU-accelerated primal heuristics to quickly find high-quality feasible solutions.

Who it’s for

It is intended for developers and researchers working on large-scale optimization, logistics (routing), and mathematical programming who need the speed of GPU acceleration and the ability to deploy across hybrid or multi-cloud environments.

Highlights

  • Broad Optimization Support: Handles LP, QP, VRP, and beta support for QCQP, SOCP, and MIP.
  • Massive Scale: Capable of solving LPs with millions of variables and constraints.
  • Flexible Integration: Provides Python, C, and Server APIs, including an algebraic modeling Python API for building constraints and objectives.
  • Agent-Ready: Includes open-source agent skills to integrate optimization into agent-first workflows.

Related

  • Project
    NVIDIA/cuopt-examples

    A collection of examples and notebooks demonstrating how to use NVIDIA cuOpt, a GPU-accelerated engine for solving complex decision optimization problems like vehicle routing and linear programming.

  • Project
    NVIDIA-AI-Blueprints/portfolio-optimization

    A GPU-accelerated portfolio optimization framework that uses NVIDIA cuOpt and RAPIDS cuML to speed up scenario generation and Mean-CVaR risk modeling.

  • Project
    NVIDIA/Model-Optimizer

    NVIDIA Model Optimizer (ModelOpt) is an open‑source Python library that lets you apply quantization, pruning, NAS, distillation, speculative decoding, and sparsity to Hugging Face, PyTorch, or ONNX models and export them for fast inference with NVIDIA runtimes like TensorRT‑LLM, vLLM, and SGLang.

  • Project
    neka-nat/cupoch

    A GPU-accelerated library for rapid 3D data processing in robotics, providing high-performance point cloud algorithms, collision avoidance, and path planning.

  • Project
    NVIDIA/cuml

    A GPU-accelerated machine learning library that provides scikit-learn-style APIs and a transparent acceleration module to speed up ML workloads on NVIDIA GPUs.