Ives Street is excited to announce the release of our travel demand modeling software, Connectome, under a license that makes it free to use for public agencies, nonprofits, and educational institutions. We’ve been developing Connectome for several years, since before Ives Street existed, and we plan to continue releasing updates and improvements for a long time to come. Connectome is multimodal travel demand modeling software that calibrates to any region in the United States from open data, runs scenarios in hours, and compares highway, transit, bicycle, and land-use projects on one scale.
Today we’re publishing the full source code of Connectome v0.8.0 at github.com/Ives-Street/Connectome. Anyone can read it, run it, and audit it. Public agencies, nonprofits, and educational institutions can use it, free, for real planning and decisionmaking, including client work.
Connectome isn’t necessarily an agency’s travel demand model of record. But it can be calibrated, by us or by agency staff, to model-of-record standards. Ives Street works directly with agencies, and we also join teams with firms specialized in planning or engineering: we can license the software or run it on your behalf.
This release is a production beta. The code is stable and complete enough that we run client studies on it, and it comes with a complete example study of Chittenden County, Vermont. It’s a beta because there’s no installer or package yet, releases are frequent and irregular, and interfaces and file formats still change between versions without a deprecation cycle. It runs on Linux from the command line.
What sets Connectome apart?
Connectome is engineered from the ground up for real-world planning and decisionmaking. Three aspects of the software make sure that it’s not only meaningful but also practical: it’s relatively quick to set up and run, it includes detailed analysis of all modes of travel, and it can assess projects in all modes and even in land-use under the same unified metric of economic benefit.
Connectome is fast enough that it can be integrated into iterative planning processes, not just used to evaluate a plan that’s already been written. Evaluating a new scenario is a matter of minutes in a small city, or a few hours in a multi-million-person metro area, on consumer-grade hardware. (Runtimes vary by the scope and resolution of the study, parameters that can be adjusted with ease.) Calibrating Connectome to a new region is also straightforward: we’ve built in pipelines to do a full regional calibration nearly automatically, using open data sources such as the US Census and OpenStreetMap. Of course, region-specific data is almost always preferable, and Connectome can easily be calibrated to an existing regional household survey.
Roads, traffic, and parking are crucial to transportation outcomes, and so are land use, public transport service, bicycle infrastructure, and walkability. Connectome represents all of them in detail. We’ve included factors that are rarely seen in trip-based models, such as the availability of parking garages, the safety of the walking environment, and the variation in public transport service over an hour-long window. Modeling all these factors and many others requires detailed parameters and occasional assumptions. We publish these parameters and we make it easy for users to adjust them for sensitivity testing or localization.
Transportation decisions are always a question of tradeoffs. A roadway widening might ease traffic on a stretch of highway even as it induces travel that will exacerbate congestion on the arterials that feed it. A bicycle lane might make it more comfortable for cyclists to ride down a street but harder for motorists to park. Denser development might put amenities within walking distance of more residents while also causing traffic that slows down buses passing through.
Connectome lets you understand these tradeoffs. In addition to measuring conventional indicators like vehicle-hours of delay or vehicle-miles traveled, Connectome is purpose-built to calculate the metric that transportation economics generally regards as the most theoretically robust metric of consumer surplus value created by a transportation project: logsum utility. Logsum utility captures the total benefit that people get from their ability to reach all the destinations they might want to visit. An increase in logsum utility can be expressed as an equivalent number of minutes or dollars saved, so even though it’s mathematically nuanced it’s easy to communicate results. Connectome lets you see how this utility will be distributed: sometimes a project might increase utility for a particular demographic or geographic group while decreasing access for another. But the best thing about logsum utility is that it’s the same metric for all modes of transportation, and so you can use it to do a single cost-benefit analysis that compares directly between a roadway widening, a light rail, a bicycle lane program, or any other transportation or land use project you can imagine.
You can read more about Connectome’s methodology and technology in the documentation.
These three characteristics mean that Connectome is a practical tool for many different parts of transportation and land-use planning. It’s convenient enough to be useful for organizations that previously lacked the resources for travel demand modeling. It’s fast enough to use to evaluate dozens of scenarios or for a quick back-and-forth between planners and modelers. It can do detailed assessments of all different modes of travel, all on the same performance metric. You can use Connectome for long-range planning, for corridor studies, for public transportation service planning, for exploring long-range planning scenarios, for project prioritization, and for anything else you can think of. And on top of all that, it’s free.
The Business Source License
We released Connectome under the Business Source License v.1.1. Anyone at all can download the entire Connectome source code and run it for their own use in research or testing. Public agencies, non-profit organizations, and educational institutions can also use Connectome in production use for real-world planning and decisionmaking and even for client work. (The full license is available at https://github.com/Ives-Street/Connectome/blob/main/LICENSE, this section is only a summary.)
There’s only one restriction. This license does not allow for-profit firms to use Connectome in production. Firms wishing to use Connectome in production should contact us to discuss an agreement. This restriction applies even if the funding originally comes from a public source, and so agencies wishing to implement a Connectome model have four options: they can do it in-house, they can work with a nonprofit or educational group, they can consult with Ives Street, or they can consult with a firm that licenses the code.
The Business Source License covers the full code for the Connectome model itself, the exact software that we use to serve our clients. It does not include the interactive visualizations that we’re also developing to display the results of a model run in a user’s web browser. That code remains proprietary to Ives Street; contact us if you’re interested in using it. Other than the visualization, the source-available license covers the entire codebase that Ives Street uses in production, and we’ll continue to maintain it with the improvements we make internally. We have a lot of plans to improve Connectome – to make it run more quickly, to model pedestrian and bicycle environments with more nuance, possibly to explore quasi-dynamic traffic assignment – and all of that work will be added to the repository as soon as it’s operational.
Why did we release Connectome under a source-available license? This is uncommon in the field of travel demand modeling. Legacy modeling software is usually proprietary and closed-source, and open-source modeling software is generally supported either by a public consortium or by academic research groups. What led Ives Street, a private firm, to give away our intellectual property so freely?
Connectome is open because in this field, transparency matters. We want public agencies to be able to understand how Connectome works before they adopt it. We want prospective users to scrutinize the code, the parameters, the assumptions, before they rely on it to make decisions. (Of course we hope they’ll let us know if they have any concerns or suggestions!) And we want agencies to have the option of making their model free and available to the public, the way that Boston’s CTPS and the Bay Area’s MTC do. Travel demand models are used for some very important decisions, and those decisions should be made with technical information as available as possible.
Getting Started
As soon as you download Connectome and set up a Python environment, you can run it. The repository ships with a complete example study of Chittenden County, Vermont: Burlington and suburbs, about 150,000 people in 283 zones. Once the environment is built, a single command sets up the region, calibrates it, and runs three scenarios. It takes about half an hour, peaks under 6 GB of memory, and needs no API keys and no network connection. When it finishes you’ll get disaggregated results exported as a GeoPackage. The package includes reference numbers, operator documentation, and a walkthrough of the example study.
Connectome v0.8.0 has limitations, and the methodological documentation lists them in detail. Connectome models a single analysis hour, usually the morning peak, and reports per-hour volumes. Daily expansion uses fixed factors, and there’s no time-of-day choice. It’s a trip-based model, not an activity-based one, so it doesn’t represent tours or scheduling. Traffic assignment is static user equilibrium, which means no queue spillback and no intersection interaction. Addressing these limitations is on our to-do list, and we plan to update the repository with frequent improvements.
If you work at a public agency, please get in touch: we’ll calibrate a demo to your region and a planning question you’re working on. If you’re at a planning or engineering firm, we’d like to hear from you too. We offer services as a modeling subconsultant and we license Connectome for production use. If you’re a researcher or a student, it’s yours to use; we’d appreciate a citation and we’d especially appreciate your feedback and bug reports. Please get in touch at dtr@ives.st.
By the way, we’ll be at the AMPO Annual Conference in Sacramento next week, September 13th-17th, 2026, and then the MobilityData Summit in Montreal, September 21st-23rd. If you’ll be at either one, let us know.