Bring us a challenging optimization problem. Our team will review your model, run it against suitable solvers, and explain the results in a written report.
We welcome GAMS and MPS models. Other formats can be discussed. The service is available to you whether or not you already use GAMS.
At no charge. No software purchase obligation.
Compare solver and hardware choices against your runtime, solution quality, and cost requirements.
Measure individual solve times and, where relevant, model preparation time.
Compare the solutions achieved within the available time, including feasibility, objective values, and optimality gaps where applicable.
Examine reliability and consistency across model instances and repeated runs, including unsuccessful runs and numerical difficulties. Assess how performance varies with small changes in the computing environment.
Weigh performance against license prices and compute costs using explicit assumptions about your intended usage and deployment.
Solver price comparisons use GAMS licensing and pricing assumptions. Prices and terms offered directly by solver vendors may differ.
We can help you assess which hardware and resource settings suit your workload.
Compare CPU configurations, thread counts, and algorithm memory requirements.
For related model instances, compare how many can be solved within a time and resource budget using more threads per solve or more concurrent solves.
Where the selected solver and algorithm support GPU acceleration, we can assess whether to include GPU tests.
Start with a short description of your model and what is difficult about solving it.
If available, describe your current solver, software versions, hardware, and settings, and tell us whether you have existing run logs.
We review whether the service fits your model and the decision you want to make. Together, we agree on suitable solvers, the number of model instances, metrics, hardware comparisons, resource limits, and a timeline.
Memory use and other relevant measures can be included; solver tuning or model reformulation can be discussed as additional scope.
We also agree on confidentiality and the terms for use of the results before evaluation work begins.
Once confidentiality and file-transfer arrangements are agreed, share your model and required data.
We review your files and run the comparison under documented conditions, using any supplied run logs to establish a baseline.
The written report includes:
You can also arrange a 30-minute face-to-face session with our experts to discuss the findings and ask questions.
We provide this evaluation as pre-sales support to help you assess our software on your own problem and to better understand current optimization needs.
All solvers available through GAMS are eligible for evaluation, including commercial and open-source solvers. Together, we select suitable solvers for your model and the decisions you want to make. See the GAMS solver documentation for available solvers and their capabilities.
We build and maintain solver interfaces and work directly with solver developers. Decades of integration and support experience help us interpret the results and explain their practical implications.
We compare the agreed model instances under documented test conditions, using comparable resource limits. We record solver settings and explain differences in hardware or resource allocation when these are part of the comparison. The report includes unsuccessful runs and the limits of the conclusions, so you can assess the trade-offs behind our recommendations.
Tell us about your confidentiality requirements in your initial inquiry. Before you send confidential material, we agree on the arrangements for transfer, access, retention, and deletion, including any non-disclosure agreement needed for the evaluation.
Reports and evaluation results must not be published or disseminated publicly. This is a mandatory condition of the service and forms part of the evaluation terms.
Findings depend on the individual model, data, solver settings, and test environment. Results for one model cannot be extrapolated to other models or presented as a general ranking of solvers.
We can investigate all model types that can be represented in GAMS, including LP, MIP, QP, NLP, MINLP, and models with positive semidefinite (PSD) cones. Solver selection and model representation are agreed for your evaluation.
The service is intended to help with solver or hardware decisions, including models with long runtimes, large optimality gaps, or numerical difficulties. We review suitability and agree on the test workload before starting.
Our in-house compute nodes have 128 GB of memory. If your evaluation requires more memory, we need to discuss arrangements. The number of instances, solvers, and resource limits are agreed as part of the scope.
No. We welcome GAMS and MPS models. If you use another format, tell us what you have so we can discuss suitability.
No. The solver evaluation service is not available to academic users. For academic licensing and resources, see the GAMS Academic Program.
We respond to initial requests within two business days. As soon as we understand the scope of your evaluation, we will provide a timeline estimate. We agree on the scope and timeline with you before evaluation work begins.
Email our team using the prompts below. Rough answers are sufficient, and we can follow up on missing details.
Please wait until confidentiality and file-transfer arrangements are agreed before sending model files or logs.
We respond to initial requests within two business days.
Request a free evaluationEmail: evaluation@gams.com
