Wolfram Network Licensing
A flexible and cost-effective license, shared across your team
A network license lets a group of people share Wolfram products—Mathematica, System Modeler and others—from a central pool instead of licensing products for every person or machine individually. Install the software widely and pay only for a maximum number of concurrent users, controlling processes and compute processes at a given time. It is the most cost-effective way to give a group access when not everyone needs it at once.
Ready to set it up? See our setup and installation resources or contact us for a quote or sizing help.
How It Works and Why It Helps
The product is installed on as many machines as you like. A small program called MathLM, the Wolfram license manager, runs on a server and tracks how many controlling and compute processes are active at any given time. That single idea is what makes a network license flexible and cost effective.
Install broadly
Put it everywhere it's needed
Deploy Mathematica across labs, offices and shared machines without buying a license for each one. Coverage isn't the thing you pay for.
Pay for concurrency
Only simultaneous use counts
You license how many people or processes can use the product at the same time. When usage varies across a group, this covers far more people than the same number of individual licenses.
Manage centrally
One server, one place to look
MathLM hands out access as sessions start and reclaims it as they close. Administrators see usage in one place and can size the license to real demand.
Your work stays on your machines
Mathematica runs locally on each computer. MathLM manages licensing—not computation—so notebooks, data and results stay in your environment.
What's the Right Size?
A network license is sized by three things, and it measures concurrency rather than head count. Because serious use often means more than one session per person, five concurrent users can need a Network 10, not a Network 5.
Concurrent users
The people connected
The people using the license at a given time—the starting point for sizing, though the license doesn't meter people directly.
It is about live sessions, not head count: one person often runs more than one at once. Because the software installs broadly while only a set number of sessions run at once, a license can cover a group larger than its seat count when use is spread through the day.
Controlling processes
Active sessions
The number the “Network N” counts. Each session is one persistent kernel a client keeps open: a notebook, terminal, IDE workflow or agent.
A controlling process is a persistent Wolfram Language kernel a client keeps open—one kernel that successive calls share, keeping variables and history. An AI assistant inside a notebook shares that notebook's kernel, but an agent that opens its own connection through a local MCP server counts separately.
A single VS Code workflow can drive several controlling processes at once—language server, in-IDE terminal, MCP connection and notebook—so one developer can hold up to four. A “Network 10” allows ten at once: ten people running one each, five running two each or one power user consuming most of it.
Compute processes
The parallel pool
A separate shared pool for parallel computation, sized with the license (for example, 80 with a Network 10). Any job can draw on whatever is free.
A compute process is an extra kernel that runs part of a parallel computation. You draw on them with parallel code—Parallelize, ParallelTable, ParallelEvaluate—which launches subkernels for the length of a job and releases them when it finishes. The pool is shared across the whole license, so one job can use 32 while another uses 8 and the rest sit idle. See the Parallel Computing guide.
Need more? The pool expands with core extensions, a power network or Wolfram Compute Services for large or bursty workloads.
A snapshot of a network in use
One Network 10, three patterns of use.
What this shows
The same Network 10 can support very different combinations of people, agents, automation and workloads. The pool could support eight people with mostly individual sessions, a smaller team working alongside agents and automation or three people running several sessions each. Compute-process demand varies separately, from 40 to all 80 processes in these examples. Size the license for peak concurrent sessions and expected parallel computing demand rather than head count alone.
Sizing in practice
University lab
20 students, but rarely just one process each
A teaching lab has 20 machines and up to 20 students at once. Counting students suggests a Network 20, but if each keeps two sessions open (a notebook and an IDE), real demand is twice that.
Network 20
→
Network 40
Company team
5 users, two controlling processes each
About five people use Mathematica at any one time. If each keeps two controlling processes active (a notebook and an IDE), the controlling count alone points to a Network 10. Heavy computation can expand the shared compute pool on top of that.
Network 5
→
Network 10 (+ compute pool)
Resources
Setup & Installation
What Is MathLM?
Overview of the license manager and how it administers network licenses.
Install MathLM
Installation guides for Windows, macOS and Linux servers, plus upgrade steps.
Mathematica + System Modeler
Set up a single MathLM server to license both products together.
Provisioning Network Access
Step-by-step article for connecting client machines to the server.
System Administration Guide
Full guide to administering Wolfram products, including license setup and deployment.
All License Manager Topics
Browse every MathLM and license manager support article in one place.
Operate & Monitor
mathlm Command
Command-line options for the license server: MathID, timeouts, policy files and more.
monitorlm Command
Check server status and current license usage, locally or remotely.
Logging MathLM
Configure usage and activity logs in standard W3C/Apache log format.
Troubleshooting MathLM
Debug client connections, password-file errors and process-limit messages.
Downloads & Accounts
FAQ
How do we install and set up a network license?
Install MathLM on a stable, always-on server (Windows, macOS or Linux), install the Wolfram products on your client machines, then point the clients at the server. MathLM must stay running for clients to launch. The full, version-specific steps for installing MathLM are maintained in Wolfram documentation.
Where do people on the network download the software?
It depends on the role. Administrators manage licensing, activation keys and downloads through Wolfram Account, which end users usually can't access. End users don't need it: from the Wolfram Download Center they can download the latest Wolfram application (used for Mathematica, Wolfram Finance Platform and others) or the separate System Modeler application, then point the installed product at your MathLM server.
How is a network license different from a single-user license?
A single-user license is tied to one specific machine, no matter how many people use that machine, so it fits a personal laptop or a shared lab computer equally well. A network (concurrent) license instead lets the software be installed across many machines and draws from a central pool sized by how many are in use at once, which suits labs, classrooms and shared workstations used across a group. Network-licensed software needs network access to a server running MathLM.
When should I get single-user licenses instead of a network license?
Because a network license needs an active connection to MathLM and grants access first come, first served, a mix of the two sometimes works best. You might use a network license for student use and dedicated single-user licenses for faculty offices. Or you might give a primary user who travels and may not have consistent internet a single-user license and cover most on-premises use with a network license.
On how many machines can we install it?
You can install it on as many machines as you like. There is no cap on installations; MathLM simply controls how many instances can run at the same time, so what you license is simultaneous use, not the number of installs.
What does "concurrent use" mean in practice?
It is the number of sessions or processes that can run at the same time, granted first come, first served. With a Network 5, different machines can rotate through the day as long as no more than five controlling processes are in use at once. If someone launches when all are in use, they may see a "not enough licenses available, try again later" message.
Can we see who is using the license?
Yes. The monitorlm tool gives a live view of server status and which processes are in use, locally or remotely, so you can tell whether users are hitting the limit. MathLM also writes usage logs, so you can review how many processes were in use over time and size the license correctly.
Where does our data go?
With a network license, you are not sending your users' work to Wolfram. MathLM runs on a server your organization operates, and clients contact that server for license access, not Wolfram. Computation runs on client machines, so notebooks, data and results stay in your environment unless users choose other Wolfram services. Usage logs are written to your server and controlled by your administrators. Wolfram follows GDPR and CCPA, and you can read more about how Wolfram approaches trust and data security. For teams that want a fully self-contained deployment, Wolfram offers Enterprise Private Cloud.
How does Wolfram generate our license password?
Your license password (the mathpass file) is generated for your server's MathID, a value derived from the machine that runs MathLM. You provide that MathID when you obtain or move a license so Wolfram can create the correct password file. It identifies the license server, not your users or their work.
Can people work offline or off the network?
We offer various licensing options for offline and off-network use. Please contact us if you wish to discuss details.
Does the same server handle System Modeler?
Yes. A single MathLM installation can manage network licenses for both Mathematica and System Modeler.
Does the license server need to be powerful?
No. MathLM hands out and tracks licenses; it does not run computations. All it needs is a stable, always-on Windows, macOS or Linux server with a reliable network connection.
Where can a network license be used?
Each network license has a standard usage zone, and most organizations upgrade to global access so anyone in the organization can use it regardless of physical location.
Do we need more than one license server?
MathLM can only serve clients that can reach it within a trusted network environment. If your network is partitioned into segments that cannot all reach one server, you will need a separate license server in each partition.
Can a network license be used for HPC computing?
Yes. Because a network license is measured by controlling and compute processes, not by head count, a common use case is a single user running Mathematica on an HPC cluster. Even though it is only one user, that user may draw on the entire pool of compute processes for a single job.
Can I add compute processes without paying for more users or controlling processes?
Yes. The shared compute pool can be expanded on its own through core extensions or a power network without increasing the number of concurrent users or controlling processes. Contact us if you are not sure how much compute you need.
Does MathLM support Mathematica installed on different operating systems?
Yes. Mathematica can be installed across platforms such as macOS, Linux and Windows, and MathLM connects to those clients regardless of platform.
Can MathLM or Mathematica run on cloud infrastructure?
It depends. Please contact us so our sales team can review your infrastructure and confirm the details.
What is the smallest network configuration available?
The smallest configuration sold is Network 2, a network license with two concurrent seats (controlling processes).
Need a quote or sizing help?
For proposals, pricing and help choosing the right concurrent count for your organization, contact us.