I've had firewalls which have run for over eight years with no reboots. That was mostly because the applications which sent traffic through them had conflicting windows for potentially-disruptive changes, and we could never get a window from all of them at the same time.
Long uptime gives me hives because it means you haven't recently tested to be sure the firewall can come back up after losing power. I've had a few datacenter-wide power outages (fire in the power distribution room, state superconducting grid outages, etc.). Every time, some system which has been too critical to maintain hasn't come back up. Most of the times, they're not one of my systems, but a few have been.
Today, we install a jumbo every 180 days at most, and I'm working towards every 90 days. For years, the first response on every ticket we opened was "You're on an old version. Jumbos include a lot of fixes. Try updating the jumbo and tell us if the issue is still present." Since we've gotten serious about more frequent updates, we get that a lot less, and every ticket spends a week less dealing with that kind of boilerplate. The work to let us update more frequently (e.g, finding and eliminating differences between cluster members) has also led to much greater overall reliability on our firewalls.