# Ways to Get Started With NVIDIA cuOpt

Streamline your optimization problems from data to decisions.

Try

### Google Colab

Experience NVIDIA® cuOpt™ on Google Colab for GPU-accelerated decision optimization for a diverse set of use cases available for rapid exploration and experimentation.

[Try Now](https://colab.research.google.com/github/NVIDIA/cuopt-examples/blob/cuopt_examples_launcher/cuopt_examples_launcher.ipynb)

Try

### NVIDIA API Catalog

Experience NVIDIA cuOpt for accelerated decision optimization of an interactive vehicle routing problem (VRP) example through an API interface.

[Try Now](https://build.nvidia.com/nvidia/nvidia-cuopt)

Develop

### GitHub

NVIDIA cuOpt is available as open-source software on [GitHub](https://github.com/NVIDIA/cuopt), [PIP](https://github.com/NVIDIA/cuopt/blob/branch-25.10/README.md), [Docker](https://hub.docker.com/r/nvidia/cuopt), [Conda](https://anaconda.org/rapidsai-nightly/cuopt-server), and [NVIDIA NGC](https://catalog.ngc.nvidia.com/orgs/nvidia/teams/cuopt/containers/cuopt?version=25.8.0-cuda12.8-py3.12)™. Also accessible via third-party integrations with [AMPL,](https://dev.ampl.com/solvers/cuopt/index.html) [CVXPY,](https://www.cvxpy.org/install/) [PuLP,](https://coin-or.github.io/pulp/) [GAMSPy](https://www.gams.com/blog/2025/09/gpu-accelerated-optimization-with-gams-and-nvidia-cuopt/), and [JuMP](https://jump.dev/JuMP.jl/stable/packages/cuOpt/).

[Access Code](https://github.com/NVIDIA/cuopt)

Deploy

### NVIDIA AI Enterprise

Get support for cuOpt with [NVIDIA AI Enterprise](https://www.nvidia.com/en-us/data-center/products/ai-enterprise.md).

[Learn More](https://www.nvidia.com/en-us/data-center/products/ai-enterprise.md)

|  |  |  |  |  |
| --- | --- | --- | --- | --- |
| Features | Try Google Colab [Try Now](https://colab.research.google.com/github/NVIDIA/cuopt-examples/blob/cuopt_examples_launcher/cuopt_examples_launcher.ipynb) | Try NVIDIA API Catalog [Try Now](https://build.nvidia.com/nvidia/nvidia-cuopt) | Develop GitHub [Access Code](https://github.com/NVIDIA/cuopt) | Deploy NVIDIA AI Enterprise [Learn More](https://www.nvidia.com/en-us/data-center/products/ai-enterprise/?ncid=pa-srch-goog-157409-prsp) |
| NVIDIA cuOpt microservice |  |  |  |  |
| * Linear programming (LP) |  |  |  |  |
| * Mixed-integer programming (MIP) |  |  |  |  |
| * Vehicle routing problem (VRP) |  |  |  |  |
| Business-standard support, including: |  |  |  |  |
| * Unlimited technical support cases accepted via the customer portal 24/7 |  |  |  |  |
| * Escalation support during local business hours (9 a.m.–5 p.m., Monday–Friday) |  |  |  |  |
| * Timely resolution provided by NVIDIA experts and engineers |  |  |  |  |
| * Security fixes and priority notifications |  |  |  |  |
| * Up to three years of support for designated branches |  |  |  |  |
|  |  |  |  |  |
| --- | --- | --- | --- | --- |
|  | [Try Now](https://colab.research.google.com/github/NVIDIA/cuopt-examples/blob/cuopt_examples_launcher/cuopt_examples_launcher.ipynb) | [Try Now](https://build.nvidia.com/nvidia/nvidia-cuopt) | [Access Code](https://github.com/NVIDIA/cuopt) | [Learn More](https://www.nvidia.com/en-us/data-center/products/ai-enterprise/?ncid=pa-srch-goog-157409-prsp) |

| "<h3 class=""h--smallest"">Features</h3>" | <h5 class="h--smallest">Try</h5><h2 class="h--smaller">NVIDIA API Catalog</h2><p>For individuals looking to experience cuOpt with sample data via API and UI-based demos for free.</p><div class="nv-button button nv-button-text nv-button-caret nv-button-standard button-left button-lap-left button-tab-left button-mob-left"> <a class="btn-content btncta" rel="noopener noreferrer" href="/content/nvidiaGDC/zz/en\_ZZ/launchpad/ai/route-optimization/"> <span class="btn-text">Try Now</span> </a> </div> | <h5 class="h--smallest">Experience</h5><h2 class="h--smaller">NVIDIA LaunchPad</h2><p>For enterprises looking to try cuOpt before purchasing for production.</p><div class="nv-button button nv-button-text nv-button-caret nv-button-standard button-left button-lap-left button-tab-left button-mob-left"> <a class="btn-content btncta" rel="noopener noreferrer" href="/content/nvidiaGDC/zz/en\_ZZ/launchpad/ai/route-optimization/"> <span class="btn-text">Access Lab</span> </a> </div> | <h5 class="h--smallest">Deploy</h5><h2 class="h--smaller">NVIDIA AI Enterprise</h2><p>For enterprises looking to try cuOpt before purchasing for production.</p><div class="nv-button button nv-button-text nv-button-caret nv-button-standard button-left button-lap-left button-tab-left button-mob-left"> <a class="btn-content btncta" rel="noopener noreferrer" href="/content/nvidiaGDC/zz/en\_ZZ/launchpad/ai/route-optimization/"> <span class="btn-text">Request a 90-Day Licence</span> </a> </div> |
| --- | --- | --- | --- |
| <div>NVIDIA® cuOpt™ microservice <ul> <li>Vehicle routing problem</li> <li>Pickup and delivery</li> <li>Traveling salesperson</li> </ul> </div> | <span class="fas fa-check grey">&nbsp;</span> | <span class="fas fa-check grey">&nbsp;</span> | <span class="fas fa-check grey">&nbsp;</span> |
| Scale | Up to 1,000 locations | Up to 10,000~15,000 locations | Up to 10,000~15,000 locations |
| AI workflow for route optimization |  | <span class="fas fa-check grey">&nbsp;</span> | <span class="fas fa-check grey">&nbsp;</span> |
| Workload and infrastructure management features |  | <span class="fas fa-check grey">&nbsp;</span> | <span class="fas fa-check grey">&nbsp;</span> |
| <div>Business-standard support, including: <ul> <li>Unlimited technical support cases accepted via the customer portal 24/7</li> <li>Escalation support during local business hours (9:00 a.m.–5:00 p.m., Monday–Friday)</li> <li>Timely resolution provided by NVIDIA experts and engineers</li> <li>Security fixes and priority notifications</li> <li>Up to three years support for designated branches</li> </ul> </div> |  |  | <span class="fas fa-check grey">&nbsp;</span> |
|  | <div class="nv-button button nv-button-text nv-button-caret nv-button-standard button-left button-lap-left button-tab-left button-mob-left"> <a class="btn-content btncta" rel="noopener noreferrer" href="/content/nvidiaGDC/zz/en\_ZZ/launchpad/ai/route-optimization/"> <span class="btn-text">Try Now</span> </a> </div> | <div class="nv-button button nv-button-text nv-button-caret nv-button-standard button-left button-lap-left button-tab-left button-mob-left"> <a class="btn-content btncta" rel="noopener noreferrer" href="/content/nvidiaGDC/zz/en\_ZZ/launchpad/ai/route-optimization/"> <span class="btn-text">Access Lab</span> </a> </div> | <div class="nv-button button nv-button-text nv-button-caret nv-button-standard button-left button-lap-left button-tab-left button-mob-left"> <a class="btn-content btncta" rel="noopener noreferrer" href="/content/nvidiaGDC/zz/en\_ZZ/launchpad/ai/route-optimization/"> <span class="btn-text">Request a 90-Day Licence</span> </a> </div> |

## Resources

### For Developers

Discover a detailed guide on the many ways you can get started with NVIDIA cuOpt.

[Start Developing](https://developer.nvidia.com/blog/accelerate-decision-optimization-using-open-source-nvidia-cuopt/)

### Documentation

Access a comprehensive collection of guides, manuals, tutorials, and other resources in the cuOpt Documentation Hub.

[Explore Docs](https://docs.nvidia.com/cuopt/)

### Community

Engage with the cuOpt community—ask how-to questions, share best practices, connect with developers, and contribute by reporting bugs.

[Explore Forum](https://forums.developer.nvidia.com/c/ai-data-science/nvidia-cuopt/514)

## FAQs

### What is NVIDIA cuOpt?

[NVIDIA cuOpt](https://www.nvidia.com/en-us/ai-data-science/products/cuopt.md) 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](https://www.nvidia.com/en-us/glossary/mixed-integer-programming.md) ([MILP](https://miplib.zib.de/instance_details_bts4-cta.html)) / 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

### What are NVIDIA cuOpt use cases?

NVIDIA cuOpt accelerates real-time decision-making across industries, enabling faster, smarter optimization for large-scale workflows such as:

* [Supply Chain Management](https://www.nvidia.com/en-us/ai-data-science/products/cuopt/#:~:text=Portfolio%20Optimization-,Supply%20Chain%20Management,-Optimizing%20resource%20allocation): Streamline sourcing, production, and delivery to reduce costs and improve efficiency across the value chain.
* [Fleet Management](https://www.nvidia.com/content/nvidiaGDC/us/en_US/ai-data-science/products/cuopt/#:~:text=Fleet%20Management,-Efficient%20scheduling%20and): Optimize routes and schedules for large-scale vehicle fleets to reduce fuel use, idle time, and operational overhead.
* [Last-Mile Delivery](https://www.nvidia.com/en-us/ai-data-science/products/cuopt/#:~:text=Last%2DMile%20Delivery,-Efficiently%20dispatching%20truck): Adapt routes in real time to meet tight delivery windows and increase customer satisfaction.
* [Field Dispatch](https://www.nvidia.com/en-us/ai-data-science/products/cuopt/#:~:text=Field%20Dispatch,-Effective%20field%20dispatch): Efficiently assign mobile workers or technicians based on skills, location, and priority.
* [Job Scheduling Optimization](https://www.nvidia.com/en-us/ai-data-science/products/cuopt/#:~:text=Job%20Scheduling%20Optimization,-Job%20scheduling%20is): Maximize resource utilization by sequencing tasks across people, machines, or shifts.
* [Portfolio Optimization](https://www.nvidia.com/en-us/ai-data-science/products/cuopt/#:~:text=Last-Mile%20Delivery,-Efficiently%20dispatching%20truck): Balance risk and return by optimizing capital allocation across financial assets under multiple constraints.

### Is NVIDIA cuOpt open source?

Yes—[NVIDIA cuOpt](https://www.nvidia.com/en-us/ai-data-science/products/cuopt.md) is open source and available on [GitHub](https://github.com/NVIDIA/cuopt) under the [Apache 2.0 license](https://www.apache.org/licenses/LICENSE-2.0). 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](https://www.nvidia.com/en-us/data-center/products/ai-enterprise.md), making it ideal for production use in industry, research, and academia.

### Do I need a GPU to use cuOpt?

Yes—[NVIDIA cuOpt](https://www.nvidia.com/en-us/ai-data-science/products/cuopt.md) 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](https://docs.nvidia.com/cuopt/user-guide/latest/system-requirements.html) 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.

### Does cuOpt use multiple GPUs?

Yes—[NVIDIA cuOpt](https://www.nvidia.com/en-us/ai-data-science/products/cuopt.md) 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](https://docs.nvidia.com/cuopt/user-guide/latest/sh-server-overview.html#step-4-running-cuopt).

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.

### What’s the best way to get started with cuOpt?

There are a few ways users can try cuOpt.

* [Google Colab](https://colab.research.google.com/github/NVIDIA/cuopt-examples/blob/cuopt_examples_launcher/cuopt_examples_launcher.ipynb): Explore ready-to-run example notebooks.
* [Brev](https://brev.nvidia.com/launchable/deploy/now?launchableID=env-2qIG6yjGKDtdMSjXHcuZX12mDNJ): Launch GPU-accelerated cuOpt cloud instances out of the box with no setup required.
* [NVIDIA NIM](https://build.nvidia.com/nvidia/nvidia-cuopt): Try cuOpt using sample data through both API and UI-based demos. It’s ideal for those new to optimization who want a hands-on introduction.
* [NVIDIA cuOpt GitHub](https://github.com/NVIDIA/cuopt): Access resources, examples, and documentation to integrate cuOpt GPU-accelerated optimization into your applications.
* [NVIDIA AI Enterprise](https://www.nvidia.com/en-us/data-center/products/ai-enterprise.md): Get enterprise support for cuOpt production deployments.

### What is the NVIDIA API catalog?

The [NVIDIA API catalog](https://build.nvidia.com/explore/discover) 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.

### What is the NVIDIA cuOpt GitHub?

[NVIDIA cuOpt GitHub](https://github.com/NVIDIA/cuOpt-Resources) is a resource hub providing examples, documentation, and tools to help developers leverage [NVIDIA cuOpt](https://www.nvidia.com/en-us/ai-data-science/products/cuopt.md), a GPU-accelerated engine for decision optimization. cuOpt is available as open-source software on GitHub, allowing developers to access the latest features, build from source, and customize it for maximum flexibility.

### What is NVIDIA AI Enterprise?

[NVIDIA AI Enterprise](https://www.nvidia.com/en-us/data-center/products/ai-enterprise.md) 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.

### Can cuOpt be used for portfolio optimization and resource allocation?

By balancing risk and return, mixed-integer programming (MIP) drives portfolio optimization and risk management.

### Can cuOpt be used for supply chain optimization?

Mixed-integer programming and linear programming revolutionize logistics by managing inventory and production schedules more efficiently. NVIDIA has demonstrated how an [AI planner](https://developer.nvidia.com/blog/building-an-ai-agent-for-supply-chain-optimization-with-nvidia-nim-and-cuopt/) —an agent built with [NVIDIA NIM™](https://www.nvidia.com/en-us/ai.md)—empowers business operations teams to chat with their supply chain data—enabling large-scale manufacturing and delivery with unprecedented efficiency. The [AI planner](https://developer.nvidia.com/blog/building-an-ai-agent-for-supply-chain-optimization-with-nvidia-nim-and-cuopt/) use case example leverages:

* A large language model (LLM) NIM to understand planners’ intentions and orchestrate other models
* [NVIDIA NeMo™](https://www.nvidia.com/en-us/ai-data-science/products/nemo.md) [Retriever](https://developer.nvidia.com/nemo-retriever?ncid=no-ncid) to connect the LLM to proprietary data
* cuOpt for decision optimization, ensuring the most efficient planning and execution

### How much does cuOpt cost?

cuOpt is available as an open-source solution. For production deployments, [NVIDIA AI Enterprise](https://www.nvidia.com/en-us/data-center/products/ai-enterprise.md) offers enhanced security, reliability, and accelerated time-to-value, along with enterprise-class support. The full pricing and licensing details can be found [here](https://docs.nvidia.com/ai-enterprise/planning-resource/licensing-guide/latest/index.html).

### How can I get in contact with someone to learn more about cuOpt?

Fill out [this form](https://www.nvidia.com/en-us/data-center/products/ai-enterprise/contact-sales.md) to be contacted by an NVIDIA Sales Representative.