Ways to Get Started With NVIDIA cuOpt
Use the right tools and technologies to efficiently take decision optimization projects from development to production.
Features |
Try NVIDIA API Catalog |
Develop GitHub |
Deploy NVIDIA AI Enterprise |
| NVIDIA cuOpt microservice | |||
|
|||
|
|||
|
|||
|
|||
|
|||
| Scale | |||
|
100M constraints 100M variables 2B non-zeros in the constraint matrix |
||
|
1–5M constraints 1–5M variables | ||
|
Up to 1,000 locations | Up to 10,000–15,000 locations | Up to 10,000–15,000 locations |
| Zero-code model support for popular modeling tools | |||
| Workload and infrastructure management features | |||
| Business-standard support, including: | |||
|
|||
|
|||
|
|||
|
|||
|
|||
Resources
FAQs
NVIDIA cuOpt is an open-source GPU-optimized software for decision optimization, delivering world-record-breaking speed and efficiency. It enables organizations to compute the most optimal plans near real time and at scale, outperforming traditional vehicle routing problems (VRP) solutions and matching the performance of mixed-integer linear programming (MILP) / linear programming (LP) solvers.
With cuOpt, businesses can:
- Assess and manage risk across complex financial instruments
- Simulate market fluctuations to enhance risk mitigation strategies
- Accelerate finding optimal solutions for large LPs and MILPs
- Solve key optimization challenges in seconds, from LP to MIP and VRP
- Reduce costs by optimizing routes and resource allocation
- Minimize miles driven and carbon emissions for sustainability
- Accelerate delivery times without compromising quality
- Increase jobs per dispatch for higher efficiency
- Re-optimize in real time to adapt to changing conditions
NVIDIA cuOpt accelerates real-time decision-making across industries, enabling faster, smarter optimization for large-scale workflows such as:
- Supply Chain Management: Streamline sourcing, production, and delivery to reduce costs and improve efficiency across the value chain.
- Fleet Management: Optimize routes and schedules for large-scale vehicle fleets to reduce fuel use, idle time, and operational overhead.
- Last-Mile Delivery: Adapt routes in real time to meet tight delivery windows and increase customer satisfaction.
- Field Dispatch: Efficiently assign mobile workers or technicians based on skills, location, and priority.
- Job Scheduling Optimization: Maximize resource utilization by sequencing tasks across people, machines, or shifts.
- Portfolio Optimization: Balance risk and return by optimizing capital allocation across financial assets under multiple constraints.
Yes—NVIDIA cuOpt is open source and available on GitHub under the Apache 2.0 license. This gives developers and researchers full flexibility to use, modify and integrate cuOpt freely, without legal uncertainty. cuOpt uniquely combines enterprise-grade performance with open-source accessibility. It accelerates both commercial and open-source solvers, offering unmatched technical advantages in speed, scalability, and solution quality—while enabling seamless integration into existing optimization workflows. cuOpt also provides a path to enterprise support through NVIDIA AI Enterprise, making it ideal for production use in industry, research, and academia.
Yes—NVIDIA cuOpt is an open-source, GPU-accelerated solver for decision optimization and a compatible GPU is required to run it. Please refer to the system requirements for supported GPU specification. You can either launch cuOpt on a cloud instance with a supported GPU or run it locally if your machine meets the necessary requirements.
Yes—NVIDIA cuOpt supports running one solver process per GPU, allowing you to configure and scale across multiple GPUs. Incoming requests are distributed in a round-robin fashion. More details can be found in the cuOpt Overview.
However, cuOpt does NOT currently support using multiple GPUs to solve a single optimization problem, nor does it support oversubscribing a single GPU with multiple concurrent solver instances.
There are a few ways users can try cuOpt.
- NVIDIA API catalog: This open-access platform allows users to interactively try cuOpt using sample data through both API and UI-based demos. It’s ideal for those new to optimization and want a hands-on introduction.
- NVIDIA cuOpt GitHub: Repository provides resources, examples, and documentation to help developers integrate cuOpt GPU-accelerated optimization into their applications.
- NVIDIA AI Enterprise: Users can get enterprise support for cuOpt production deployments with NVIDIA AI Enterprise.
The NVIDIA API catalog provides production-ready generative AI models and an optimized inference runtime, continuously enhanced for peak performance. Delivered as microservices, these models can be easily deployed on any GPU-accelerated system using standard tools.
NVIDIA cuOpt GitHub is a resource hub providing examples, documentation, and tools to help developers leverage NVIDIA cuOpt, a GPU-accelerated AI solver for decision optimization. cuOpt will be available as open-source software at Github, allowing developers to access the latest features, build from source, and customize it for maximum flexibility.
NVIDIA AI Enterprise is an end-to-end, cloud-native software platform that accelerates data science pipelines and streamlines development and deployment of production-grade AI applications, including generative AI, computer vision, speech AI, and more. It includes best-in-class development tools, frameworks, pretrained models, and microservices for AI practitioners and reliable management capabilities for IT professionals to ensure performance, API stability, and security.
By balancing risk and return, mixed-integer programming (MIP) drives portfolio optimization and risk management.
Mixed-integer programming and linear programming revolutionize logistics by managing inventory and production schedules more efficiently. NVIDIA has demonstrated how an AI planner—an agent built with NVIDIA NIM™—empowers business operations teams to chat with their supply chain data—enabling large-scale manufacturing and delivery with unprecedented efficiency. The AI planner use case example leverages:
- A large language model (LLM) NIM to understand planners’ intentions and orchestrate other models
- NVIDIA NeMo™ Retriever to connect the LLM to proprietary data
- cuOpt for decision optimization, ensuring the most efficient planning and execution
cuOpt is available as an open-source solution. For production deployments, NVIDIA AI Enterprise offers enhanced security, reliability, and accelerated time-to-value, along with enterprise-class support. The full pricing and licensing details can be found here.
Fill out this form to be contacted by an NVIDIA Sales Representative.