Warscape
Simulated matchup ยท a model, not a forecast

๐Ÿ‡ท๐Ÿ‡ธ Serbia vs ๐Ÿณ๏ธ KosovoMilitary comparison, and what happens if they fight

If Serbia attacked Kosovo today, with no allies joining on either side, this is what the WARSCAPE engine produced: Defender holds after 365 days.

How the two militaries compare

Serbia keeps 6.3 times as many soldiers under arms. Serbia has tanks and Kosovo has none. Serbia has combat aircraft and Kosovo has none. Serbia spends 10 times as much on defence. Public, country-level figures from the data snapshot the run was pinned to (2026.09). See data sources for where each number comes from.

SerbiaKosovo
25,000Active soldiers4,000
50,000Reservists3,000
230Tanks0
300Artillery10
25Combat aircraft0
50Helicopters5
$1.5bnDefence budget$0.1bn
6.6mPopulation1.8m

What the war cost

Serbia
attacked
713
soldiers killed or wounded
3
tanks destroyed
12
aircraft lost
0
land changed hands
34%
support at home, at the end
Kosovo
defended
120
soldiers killed or wounded
0
tanks destroyed
0
aircraft lost
0
land changed hands
35%
support at home, at the end

What decided it

01
Mobilisation helped Serbia
Reservists mobilised: 128,287 vs 2,318.
02
Deployed combat power helped Serbia
Average deployed combat power: 9 vs 1 model units.
03
Air situation helped Serbia
Average share of air control for Serbia: 80%.
04
Logistics helped Kosovo
Average front-line supply efficiency: 44% vs 75%.
05
Political cohesion helped Serbia
Average cohesion: 47% vs 44%.
Think it would go differently?

This run assumed no allies, winter, full mobilization for the attacker and partial for the defender, and no nuclear weapons. Change any of that and watch it play out day by day on the map.

Other borders in this neighbourhood

All matchups โ†’
After 365 simulated days the conflict between Serbia and Kosovo ended with: Defender holds (duration limit reached). Aggregate personnel losses within the model: 713 (Serbia) and 120 (Kosovo). This page reports one seeded run of a model on public, aggregate data. It is not a prediction, and no operational information is used.

Model-generated simulation on public, country-level data. Not a prediction. No operational, classified or real-time information is used.