Gurobi 9.5

Enterprise Features and Even Better Performance

Our Latest Release

Gurobi 9.5 provides customers with new enterprise features, product enhancements, and an even faster compute engine—with impressive performance improvements across all supported problem types.

Faster Than Ever

Compared to the Gurobi 9.1. release, Gurobi 9.5 has boosted its speeds across the board.

PROBLEM TYPES OVERALL SPEED-UP (>1s) HARD MODELS (>100s)
LP
  • Concurrent
  • Primal Simplex
  • Dual Simplex
  • Barrier
  • 14%
  • 23%
  • 20%
  • 18%
  • 54%
  • 43%
  • 43%
  • 56%
MIP
  • MIP
  • Convex MIQP
  • Convex MIQCP
  • Non-Convex MIQCP
  • 15%
  • 30%
  • 33%
  • 3.0x
  • 27%
  • 68%
  • 78%
  • 7.5x

Gurobi 9.5 Launch Event

Enterprise Features

Gurobi Optimizer 9.5 also includes new features that make it well suited for enterprise use, including the following:

Seamless Docker Deployments

Run Gurobi in container environments, easily, with the Gurobi Web License Service (WLS).

User-Level Management

Manage users and assign permission levels in a centralized location, by integrating Gurobi Compute Server and Cluster Manager with LDAP repositories.

Better API Key Management

Label API keys with descriptive information in the Cluster Manager to simplify API key tracking and management.

New Features and Enhancements

Customers will discover over a dozen enhancements across the product, such as powerful new heuristics for non-convex quadratic models, norm constraints, deterministic work measures, memory limit parameters, and more user control of IIS computation, as well as improvements to callbacks and tuning.

Gurobi Optimizer

NLP Heuristic for Non-Convex Quadratic Models
We have added a non-linear barrier solver for quickly finding feasible solutions to non-convex quadratic models.

Lazy Constraints in the NoRel Heuristic
Lazy constraints are now supported in the NoRel heuristic. You will receive solution callbacks and be given the opportunity to cut off those solutions with lazy constraints.

New Callback for Exiting the NoRel Heuristic
The new Proceed callback method allows you to smoothly transition from the NoRel heuristic to the MIP search.

Web License Service (WLS)
The new Web License Service enables Gurobi on containerized environments (e.g., Docker, Kubernetes). Predefined images are available on Docker Hub.

Work Metric
We now include a deterministic work measure to enable exact reproducibility of a solver run.

Memory Limit
The new MemLimit parameter allows you to terminate optimization when the memory used by the optimization exceeds the specified value.

New Cut Type – Lift-and-Project Cuts
This release includes a new cut type: lift-and-project cuts. The new LiftProjectCuts parameter provides some control over the generation of these cuts.

New Simplex Warm Start Options
Several new options are available
for starting simplex from a previous solution. Options are chosen using the new LPWarmStart parameter.

Writing Dualized Models
The new .DUA and .DLP file formats allow you to write the dual formulation of an LP problem.

Norm General Constraint
Add norm constraints to your MIP model, with support for 0-, 1-, 2-, and infinity norms.

Universal2 Port for Mac
Supports both Apple silicon and Intel x64 processors.

IIS Callbacks
New IIS callbacks allow you to programmatically track the progress of the IIS algorithm.

SOS Encoding Options
SOS constraints can now be formulated in multiple different ways internally, which can sometimes give better performance.

Tuning Tool Enhancements
We added a number of additional controls to our tuning tool, including control over termination and control over how runtimes from multiple trials are aggregated.

Improvements in gurobipy
Type hints are now available for most of the gurobipy classes and functions. Moreover, indexing MVar and MConstr objects is now more consistent with numpy.

Additional New Parameters and Attributes
MaxVio: New attribute that gives the maximum (unscaled) constraint violation for the computed solution.
ConcurrentWinMethod: New attribute that indicates which algorithm prevailed in a concurrent LP solve.
PoolIgnore: New attribute to select variables that should be ignored when checking whether two solutions differ in the solution pool.

IIS Force Flags
You now have more control over the Irreducible Infeasible Subsystem (IIS) algorithm. Specifically, you can indicate that certain constraints or bounds should be forced into the IIS, even if the resulting subsystem is not minimal. Similarly, you can exclude constraints or bounds from the IIS.

Compute Server and Cluster Manager

Web License Service (WLS)
The new Web License Service also enables Gurobi Compute Server and Cluster Manager to run on containerized environments (e.g., Docker, Kubernetes). Predefined images are available on Docker Hub.

Integration with LDAP Repositories for Centralized User Management
LDAP support allows enterprise-grade Compute Server deployments using an industry-standard directory service for user authentication and authorization.

Amazon Web Services DocumentDB 4.0 Database Support:
The Cluster Manager now supports AWS DocumentDB 4.0 as a back-end database. This expands your options when deploying the Cluster Manager on an AWS environment.

API Key Management
The Cluster Manager supports the ability to disable/enable keys, set a description, and track API keys for improved security.

Account Management
The Cluster Manager now provides system accounts, password policies, read-only role usage and the ability to disable/enable accounts for improved user management.

Usability
The Cluster Manager now provides contextual help, settings management, permalinks, and job metrics for enhanced usability.

Upgrade Now

Commercial users with an existing 9.0 license, and all eligible Academic users can upgrade to 9.5 immediately by visiting the downloads page (while logged in). Commercial users with an older version of Gurobi should contact sales@gurobi.com to get this new release. All other users, please contact us to discuss upgrade options.

Contact Us

Thank you! The information has been submitted successfully.