[fixermark]

Mapzen seems like a good idea, but I can't help but wonder (with respect): Can a developer expect it to be here in 2026?

Seems like someone integrating against an API has the choice of "Use one from a big company with a proven business model and risk their TOS changing to have it taken away" vs. "Use one from a firm with only a few years of history and risk them going dark one day (or changing their TOS)."

In this specific scenario, I'm not sure that this solution would guarantee the developer could have avoided this failure mode.

[burritojustice]

A great question. That's why we make all the underlying source code open, including chef and docker config info for people to build it themselves if they want or need to.

https://github.com/valhalla/

We at Mapzen are explicitly designing our projects to outlast us (if necessary).

https://mapzen.com/blog/our-magna-carto

[mikeg8]

Id just like to say that I really like the design of your site, and logo in particular. Well done.

[danpat]

There's also: Project OSRM: https://www.project-osrm.org/

and GraphHopper https://graphhopper.com/

And for the UI, you can use https://github.com/perliedman/leaflet-routing-machine

all of which you can self-host using OpenstreetMap data.

[HaloZero]

You're right, it's a bit of Catch-22. But the benefits of using the open approach though is that it helps promote them which could help get them marketshare/mindshare and help bring attention/support to them. At least that's what i'm hoping would be the case.