NVIDIA cuOpt is NVIDIA’s GPU-accelerated platform for operations and logistics that enables near real-time routing optimizations, empowering developers to leverage larger data sets and faster processing, delivering new capabilities like dynamic-rerouting, new algorithms, and sub-second solver response time. NVIDIA cuOpt would easily integrate with most applications and accelerate their processing time with better accuracy.


Operations Research (OR) and logistics are incredibly compute-intensive with massive operational costs. Most problems are NP-hard, and the GPUs bring the throughput capabilities needed to fuel the most ambitious heuristics (tabu search, guided local search, ant colony) while supporting the most challenging constraints (multi-capacities, time windows, heterogeneous fleet)

NVIDIA cuOpt is exposed as a fast combinatorial optimization solver available as;

  • Server.

  • Python APIs.

The solver has been tested in Last-Mile Delivery applications from retail industries to food services. The team is currently working on expanding to factory and warehouse use cases with a focus on Autonomous Mobile Robots (AMRs).