Getting Started with VeloxQ

To use the VeloxQ SDK, you first need a VeloxQ API token, set as the VELOX_TOKEN environment variable before running SDK calls. You can obtain access to a token by contacting us via the QMZ.AI’s contact form, or by using supercomputing and research infrastructures that provide access to VeloxQ services where available (for example through partner institutions such as ACK Cyfronet AGH and related access programs).

VeloxQ is built to solve QUBO/Ising-type optimization problems, which are commonly used to model complex decision-making tasks with many interacting constraints and trade-offs. In practice, this means your team starts by expressing the problem in a compatible form—such as biases and couplings, a supported file format, or a model object from an existing optimization workflow.

The VeloxQ SDK is designed to support multiple input types, so you can begin with a small prototype or connect VeloxQ to a more mature internal pipeline without rebuilding everything from scratch. This makes it easier to move from experimentation to production in a structured way.

Once the problem is prepared, it is submitted to VeloxQ through the SDK, which manages the execution workflow using core objects such as Problems, Files, Solvers, and Jobs. This structure helps teams organize optimization runs, reuse uploaded instances, track experiments, and scale usage across projects.

You can start with a simple one-call workflow using solver.sample(...), which handles the full process in a single step (create problem file, submit job, wait for completion, and return results), or use asynchronous job submission for production systems that require queueing, monitoring, and non-blocking execution. This flexibility allows VeloxQ to fit both early-stage testing and enterprise-grade deployment models.

After execution, VeloxQ returns results as a VeloxSampleSet, which includes solution samples, energy values, and run metadata. This gives your team a practical foundation for comparing candidate solutions, evaluating optimization quality, and feeding outputs into downstream applications such as scheduling systems, planning tools, analytics dashboards, or business process automation.
 
For deeper analysis and repeatability, results can also be exported and archived in structured formats, enabling benchmarking, validation, and internal reporting across multiple runs.
 
In this way, VeloxQ supports not only solving individual optimization problems, but building a repeatable optimization workflow that works on today’s infrastructure and remains ready for future hybrid quantum-classical integration.