GeoCrossBench: paired seed × dataset bootstrap

Each lower-triangle cell shows row minus column in percentage points, a paired crossed-bootstrap 95% interval, and the bootstrap probability that the row model is higher. Select a benchmark scope below.

28 model/regime combinations 378 pairwise comparisons 276 pointwise intervals exclude zero
Rank · model/regime mean1ViT-B2Panopticon3χViT4Panopticon ❄5DINOv26ResNet507DOFA8DINOv2 ❄9iBOT10DINOv311SatlasNet12Prithvi13TerraFM14χViT ❄15TerraMind16ViT-B ❄17DOFA ❄18TerraFM ❄19ResNet50 ❄20CROMA21iBOT ❄22DINOv3 ❄23TerraMind ❄24SatlasNet ❄25CROMA ❄26Prithvi ❄27AnySat28AnySat ❄
1ViT-B45.05
2Panopticon44.85-0.20[-2.78, +3.10]P+ 41.6%
3χViT44.67-0.38[-4.25, +4.07]P+ 40.9%-0.18[-2.70, +2.14]P+ 45.6%
4Panopticon ❄44.11-0.95[-5.64, +4.05]P+ 35.5%-0.75[-4.10, +2.62]P+ 33.0%-0.57[-3.47, +2.24]P+ 35.4%
5DINOv243.81-1.25[-4.39, +2.19]P+ 23.7%-1.05[-4.73, +2.62]P+ 29.4%-0.87[-4.72, +2.41]P+ 33.5%-0.30[-3.60, +2.91]P+ 42.8%
6ResNet5043.61-1.44[-5.00, +2.38]P+ 25.5%-1.24[-4.59, +1.97]P+ 24.5%-1.06[-3.79, +1.12]P+ 21.0%-0.49[-2.87, +1.82]P+ 34.5%-0.19[-2.60, +2.55]P+ 43.1%
7DOFA41.08-3.98[-6.68, -0.89]P+ 0.5%-3.78[-6.72, -0.83]P+ 0.6%-3.60[-6.80, -0.85]P+ 0.4%-3.03[-6.16, +0.15]P+ 3.1%-2.73[-4.55, -0.84]P+ 0.2%-2.54[-4.57, -0.68]P+ 0.2%
8DINOv2 ❄41.05-4.01[-7.27, -0.51]P+ 1.0%-3.81[-7.03, -0.70]P+ 0.7%-3.63[-6.73, -0.88]P+ 0.2%-3.06[-5.40, -0.77]P+ 0.5%-2.76[-4.14, -1.31]P+ 0.0%-2.57[-4.76, -0.67]P+ 0.2%-0.03[-1.58, +1.60]P+ 48.2%
9iBOT40.95-4.11[-5.51, -2.82]P+ 0.0%-3.91[-7.65, -1.23]P+ 0.1%-3.73[-8.53, +0.20]P+ 3.5%-3.16[-8.17, +1.43]P+ 10.0%-2.86[-5.92, -0.11]P+ 2.0%-2.67[-6.56, +0.76]P+ 8.4%-0.13[-3.09, +2.26]P+ 49.0%-0.10[-3.31, +2.91]P+ 48.2%
10DINOv340.59-4.46[-5.57, -3.07]P+ 0.0%-4.26[-6.88, -1.88]P+ 0.0%-4.08[-7.65, -0.71]P+ 0.6%-3.51[-7.75, +0.71]P+ 6.1%-3.21[-5.65, -0.76]P+ 0.3%-3.02[-6.18, +0.04]P+ 2.8%-0.48[-2.77, +1.69]P+ 34.8%-0.45[-3.10, +2.22]P+ 36.5%-0.35[-1.53, +1.22]P+ 28.5%
11SatlasNet40.42-4.63[-8.84, +0.17]P+ 3.0%-4.43[-8.15, -0.50]P+ 1.0%-4.25[-7.64, -1.13]P+ 0.2%-3.69[-5.89, -1.56]P+ 0.0%-3.39[-6.41, -0.50]P+ 0.9%-3.19[-4.54, -1.64]P+ 0.0%-0.65[-3.08, +1.79]P+ 31.2%-0.63[-2.92, +1.92]P+ 30.1%-0.53[-4.32, +3.96]P+ 39.2%-0.17[-3.88, +3.80]P+ 46.6%
12Prithvi40.03-5.02[-8.10, -2.22]P+ 0.0%-4.82[-9.83, -0.25]P+ 1.9%-4.64[-10.36, +0.12]P+ 2.9%-4.07[-9.97, +1.67]P+ 8.8%-3.77[-6.80, -0.81]P+ 0.4%-3.58[-8.09, +0.68]P+ 5.4%-1.04[-4.22, +2.04]P+ 26.3%-1.01[-4.95, +2.78]P+ 30.7%-0.91[-3.50, +1.45]P+ 23.9%-0.56[-3.30, +1.91]P+ 35.2%-0.39[-5.23, +4.51]P+ 43.8%
13TerraFM39.90-5.15[-9.19, -0.88]P+ 0.5%-4.95[-9.13, -0.91]P+ 0.5%-4.77[-8.57, -1.37]P+ 0.1%-4.21[-7.20, -1.38]P+ 0.1%-3.91[-6.55, -1.35]P+ 0.1%-3.71[-5.47, -1.96]P+ 0.0%-1.17[-3.35, +0.96]P+ 15.2%-1.15[-3.16, +0.83]P+ 13.3%-1.05[-4.61, +2.74]P+ 30.1%-0.69[-4.30, +2.96]P+ 35.6%-0.52[-2.27, +1.22]P+ 27.9%-0.13[-4.37, +3.99]P+ 48.2%
14χViT ❄39.88-5.18[-10.70, +0.81]P+ 4.5%-4.98[-8.93, -1.00]P+ 0.4%-4.80[-7.23, -2.41]P+ 0.0%-4.23[-6.41, -1.82]P+ 0.0%-3.93[-7.82, +0.65]P+ 4.7%-3.74[-6.48, -0.50]P+ 1.3%-1.20[-5.12, +3.05]P+ 28.4%-1.17[-4.42, +2.68]P+ 25.7%-1.07[-6.59, +5.17]P+ 34.7%-0.72[-5.70, +4.47]P+ 39.2%-0.54[-3.65, +2.85]P+ 36.6%-0.16[-6.32, +6.67]P+ 47.0%-0.02[-3.49, +4.13]P+ 48.0%
15TerraMind38.36-6.69[-10.69, -2.50]P+ 0.0%-6.49[-11.43, -1.91]P+ 0.1%-6.31[-11.08, -2.49]P+ 0.0%-5.75[-9.68, -2.20]P+ 0.0%-5.44[-7.67, -3.21]P+ 0.0%-5.25[-7.89, -2.92]P+ 0.0%-2.71[-5.09, -0.39]P+ 0.9%-2.68[-5.02, -0.62]P+ 0.4%-2.59[-5.94, +0.91]P+ 9.4%-2.23[-5.88, +1.43]P+ 14.8%-2.06[-4.77, +0.49]P+ 6.1%-1.67[-4.98, +1.68]P+ 20.0%-1.54[-3.14, -0.00]P+ 2.5%-1.52[-6.89, +2.56]P+ 29.8%
16ViT-B ❄37.57-7.48[-11.91, -2.82]P+ 0.0%-7.28[-11.79, -2.94]P+ 0.0%-7.10[-11.38, -3.24]P+ 0.0%-6.54[-9.11, -4.08]P+ 0.0%-6.24[-8.85, -3.67]P+ 0.0%-6.04[-8.67, -3.49]P+ 0.0%-3.51[-6.12, -0.81]P+ 0.4%-3.48[-5.18, -1.78]P+ 0.0%-3.38[-7.38, +0.77]P+ 6.5%-3.02[-7.05, +1.03]P+ 9.8%-2.85[-5.31, -0.58]P+ 0.6%-2.46[-7.25, +2.23]P+ 16.7%-2.33[-4.01, -0.69]P+ 0.3%-2.31[-6.77, +1.23]P+ 13.5%-0.79[-2.71, +1.13]P+ 21.2%
17DOFA ❄36.67-8.38[-12.92, -3.42]P+ 0.0%-8.18[-11.98, -4.13]P+ 0.0%-8.00[-11.35, -4.58]P+ 0.0%-7.44[-9.52, -5.24]P+ 0.0%-7.13[-10.05, -4.27]P+ 0.0%-6.94[-9.25, -4.71]P+ 0.0%-4.40[-6.83, -1.98]P+ 0.0%-4.37[-6.32, -2.31]P+ 0.0%-4.28[-8.66, +0.55]P+ 4.2%-3.92[-8.04, +0.32]P+ 3.8%-3.75[-6.03, -1.66]P+ 0.0%-3.36[-8.42, +1.75]P+ 10.5%-3.23[-5.10, -1.36]P+ 0.1%-3.20[-6.55, -0.16]P+ 1.9%-1.69[-4.25, +1.11]P+ 11.6%-0.90[-2.75, +0.89]P+ 16.3%
18TerraFM ❄36.09-8.97[-14.28, -3.41]P+ 0.0%-8.77[-13.30, -4.12]P+ 0.0%-8.59[-12.36, -4.67]P+ 0.0%-8.02[-10.28, -5.58]P+ 0.0%-7.72[-11.23, -4.30]P+ 0.0%-7.53[-10.17, -4.81]P+ 0.0%-4.99[-8.19, -1.71]P+ 0.0%-4.96[-7.45, -2.43]P+ 0.0%-4.86[-9.85, +0.37]P+ 3.6%-4.51[-9.27, +0.31]P+ 3.7%-4.33[-6.58, -2.06]P+ 0.0%-3.95[-9.57, +1.69]P+ 9.2%-3.81[-5.82, -1.73]P+ 0.0%-3.79[-7.16, -0.68]P+ 0.7%-2.27[-5.09, +0.72]P+ 6.8%-1.48[-3.14, +0.27]P+ 4.9%-0.59[-2.21, +0.94]P+ 24.4%
19ResNet50 ❄35.99-9.06[-14.29, -3.91]P+ 0.0%-8.86[-12.54, -5.19]P+ 0.0%-8.68[-12.29, -5.26]P+ 0.0%-8.11[-10.51, -5.47]P+ 0.0%-7.81[-12.38, -3.76]P+ 0.0%-7.62[-10.66, -5.05]P+ 0.0%-5.08[-8.67, -1.96]P+ 0.0%-5.05[-8.43, -2.12]P+ 0.0%-4.95[-9.86, +0.01]P+ 2.5%-4.60[-9.49, +0.08]P+ 2.7%-4.43[-6.93, -2.21]P+ 0.0%-4.04[-10.38, +1.95]P+ 10.2%-3.91[-6.66, -1.23]P+ 0.1%-3.88[-7.48, -0.13]P+ 2.1%-2.37[-6.32, +1.44]P+ 12.3%-1.58[-4.45, +1.26]P+ 14.6%-0.68[-3.13, +1.90]P+ 29.6%-0.09[-2.54, +2.50]P+ 46.0%
20CROMA35.19-9.86[-15.45, -4.41]P+ 0.0%-9.66[-14.48, -4.96]P+ 0.0%-9.48[-13.95, -5.31]P+ 0.0%-8.92[-12.87, -5.23]P+ 0.0%-8.62[-12.81, -4.92]P+ 0.0%-8.42[-11.76, -5.35]P+ 0.0%-5.89[-9.74, -2.57]P+ 0.0%-5.86[-9.48, -2.47]P+ 0.0%-5.76[-10.77, -0.88]P+ 0.9%-5.40[-10.52, -0.69]P+ 0.9%-5.23[-8.27, -2.65]P+ 0.0%-4.84[-10.36, +0.15]P+ 3.0%-4.71[-6.91, -2.48]P+ 0.0%-4.69[-9.19, -0.48]P+ 1.4%-3.17[-6.28, -0.18]P+ 1.9%-2.38[-5.47, +0.88]P+ 7.5%-1.48[-4.33, +1.49]P+ 16.1%-0.90[-3.72, +2.03]P+ 26.2%-0.80[-4.26, +2.24]P+ 32.0%
21iBOT ❄34.23-10.83[-14.11, -7.42]P+ 0.0%-10.63[-14.47, -7.18]P+ 0.0%-10.45[-14.56, -7.00]P+ 0.0%-9.88[-13.24, -6.92]P+ 0.0%-9.58[-11.74, -7.52]P+ 0.0%-9.39[-11.73, -7.14]P+ 0.0%-6.85[-8.70, -5.01]P+ 0.0%-6.82[-8.48, -5.26]P+ 0.0%-6.72[-9.36, -3.93]P+ 0.0%-6.37[-9.22, -3.47]P+ 0.0%-6.20[-8.68, -4.06]P+ 0.0%-5.81[-9.37, -2.37]P+ 0.0%-5.67[-7.24, -4.08]P+ 0.0%-5.65[-10.52, -1.66]P+ 0.1%-4.14[-5.87, -2.40]P+ 0.0%-3.34[-5.13, -1.59]P+ 0.0%-2.45[-5.07, -0.07]P+ 2.1%-1.86[-4.60, +0.78]P+ 9.1%-1.77[-4.87, +1.27]P+ 13.3%-0.96[-4.01, +2.24]P+ 26.7%
22DINOv3 ❄33.42-11.63[-15.59, -7.32]P+ 0.0%-11.43[-15.61, -7.55]P+ 0.0%-11.25[-15.61, -7.22]P+ 0.0%-10.69[-13.55, -8.35]P+ 0.0%-10.39[-12.99, -7.97]P+ 0.0%-10.19[-12.81, -7.55]P+ 0.0%-7.65[-10.20, -5.10]P+ 0.0%-7.63[-9.54, -5.65]P+ 0.0%-7.53[-10.99, -3.81]P+ 0.0%-7.17[-10.80, -3.50]P+ 0.0%-7.00[-9.37, -4.77]P+ 0.0%-6.61[-11.22, -2.30]P+ 0.0%-6.48[-8.37, -4.60]P+ 0.0%-6.46[-11.21, -2.61]P+ 0.0%-4.94[-7.33, -2.59]P+ 0.0%-4.15[-5.70, -2.65]P+ 0.0%-3.25[-5.58, -1.17]P+ 0.1%-2.67[-5.27, -0.17]P+ 1.8%-2.57[-5.63, +0.20]P+ 3.6%-1.77[-5.06, +1.51]P+ 14.4%-0.81[-2.29, +0.82]P+ 15.9%
23TerraMind ❄32.32-12.73[-18.05, -7.20]P+ 0.0%-12.53[-17.85, -7.19]P+ 0.0%-12.35[-17.33, -7.70]P+ 0.0%-11.79[-15.43, -8.33]P+ 0.0%-11.48[-14.63, -8.27]P+ 0.0%-11.29[-14.66, -8.05]P+ 0.0%-8.75[-11.97, -5.43]P+ 0.0%-8.72[-11.35, -6.04]P+ 0.0%-8.63[-13.25, -3.83]P+ 0.0%-8.27[-13.06, -3.41]P+ 0.0%-8.10[-11.02, -5.32]P+ 0.0%-7.71[-12.63, -2.89]P+ 0.0%-7.58[-9.65, -5.64]P+ 0.0%-7.56[-12.53, -3.28]P+ 0.0%-6.04[-8.12, -4.12]P+ 0.0%-5.25[-6.87, -3.79]P+ 0.0%-4.35[-6.91, -2.01]P+ 0.0%-3.76[-5.89, -2.15]P+ 0.0%-3.67[-7.24, -0.24]P+ 1.7%-2.87[-5.75, -0.17]P+ 1.8%-1.90[-4.27, +0.49]P+ 7.1%-1.10[-3.71, +1.38]P+ 20.6%
24SatlasNet ❄29.18-15.87[-22.26, -9.49]P+ 0.0%-15.67[-20.98, -10.37]P+ 0.0%-15.49[-20.69, -10.68]P+ 0.0%-14.93[-18.46, -10.93]P+ 0.0%-14.63[-19.73, -9.69]P+ 0.0%-14.43[-19.06, -10.38]P+ 0.0%-11.89[-16.38, -7.75]P+ 0.0%-11.87[-15.70, -8.04]P+ 0.0%-11.77[-17.90, -5.53]P+ 0.0%-11.41[-17.46, -5.52]P+ 0.0%-11.24[-15.55, -7.45]P+ 0.0%-10.85[-17.99, -4.03]P+ 0.0%-10.72[-14.77, -7.28]P+ 0.0%-10.70[-15.89, -6.02]P+ 0.0%-9.18[-13.83, -4.83]P+ 0.0%-8.39[-11.51, -5.28]P+ 0.0%-7.49[-10.48, -4.77]P+ 0.0%-6.91[-10.01, -4.42]P+ 0.0%-6.81[-9.62, -4.28]P+ 0.0%-6.01[-10.88, -1.97]P+ 0.2%-5.05[-9.15, -1.07]P+ 0.5%-4.24[-7.88, -0.32]P+ 1.7%-3.14[-6.74, +0.38]P+ 4.0%
25CROMA ❄29.13-15.92[-22.11, -9.44]P+ 0.0%-15.72[-21.35, -9.87]P+ 0.0%-15.54[-20.55, -10.32]P+ 0.0%-14.98[-18.86, -10.83]P+ 0.0%-14.68[-19.31, -10.21]P+ 0.0%-14.48[-18.03, -10.57]P+ 0.0%-11.94[-16.04, -7.82]P+ 0.0%-11.92[-15.75, -8.09]P+ 0.0%-11.82[-17.43, -5.94]P+ 0.0%-11.46[-17.17, -5.64]P+ 0.0%-11.29[-14.16, -8.03]P+ 0.0%-10.90[-16.98, -5.09]P+ 0.0%-10.77[-13.24, -8.30]P+ 0.0%-10.75[-15.39, -6.13]P+ 0.0%-9.23[-12.59, -6.15]P+ 0.0%-8.44[-11.23, -5.77]P+ 0.0%-7.54[-10.52, -4.70]P+ 0.0%-6.96[-8.94, -4.95]P+ 0.0%-6.86[-10.00, -3.71]P+ 0.0%-6.06[-8.70, -3.71]P+ 0.0%-5.10[-8.34, -1.72]P+ 0.1%-4.29[-7.62, -0.95]P+ 0.5%-3.19[-5.39, -1.20]P+ 0.0%-0.05[-3.15, +3.67]P+ 46.7%
26Prithvi ❄26.40-18.66[-23.05, -14.01]P+ 0.0%-18.46[-23.13, -13.69]P+ 0.0%-18.28[-23.66, -13.03]P+ 0.0%-17.71[-22.57, -12.57]P+ 0.0%-17.41[-21.61, -13.05]P+ 0.0%-17.22[-21.30, -12.93]P+ 0.0%-14.68[-17.90, -11.36]P+ 0.0%-14.65[-18.24, -10.82]P+ 0.0%-14.55[-18.32, -10.49]P+ 0.0%-14.20[-18.30, -9.79]P+ 0.0%-14.02[-17.92, -9.95]P+ 0.0%-13.64[-18.29, -9.50]P+ 0.0%-13.50[-16.75, -10.28]P+ 0.0%-13.48[-19.76, -7.44]P+ 0.0%-11.96[-15.49, -8.53]P+ 0.0%-11.17[-14.62, -7.61]P+ 0.0%-10.28[-14.09, -6.43]P+ 0.0%-9.69[-13.61, -5.70]P+ 0.0%-9.60[-13.38, -6.06]P+ 0.0%-8.79[-12.68, -4.77]P+ 0.0%-7.83[-10.26, -5.12]P+ 0.0%-7.02[-10.05, -3.59]P+ 0.0%-5.93[-9.48, -2.24]P+ 0.1%-2.78[-6.86, +1.19]P+ 8.9%-2.73[-6.31, +0.66]P+ 7.6%
27AnySat24.50-20.56[-26.21, -14.73]P+ 0.0%-20.36[-26.69, -14.23]P+ 0.0%-20.18[-26.50, -14.05]P+ 0.0%-19.61[-25.57, -13.90]P+ 0.0%-19.31[-23.81, -14.66]P+ 0.0%-19.12[-24.06, -14.05]P+ 0.0%-16.58[-21.07, -12.15]P+ 0.0%-16.55[-20.92, -12.26]P+ 0.0%-16.45[-21.34, -11.36]P+ 0.0%-16.10[-21.28, -10.64]P+ 0.0%-15.92[-20.74, -11.13]P+ 0.0%-15.54[-20.11, -10.90]P+ 0.0%-15.40[-19.21, -11.70]P+ 0.0%-15.38[-22.30, -9.05]P+ 0.0%-13.86[-17.45, -10.35]P+ 0.0%-13.07[-17.10, -9.33]P+ 0.0%-12.17[-17.11, -7.88]P+ 0.0%-11.59[-15.89, -7.63]P+ 0.0%-11.50[-16.84, -6.54]P+ 0.0%-10.69[-15.17, -6.52]P+ 0.0%-9.73[-13.04, -6.39]P+ 0.0%-8.92[-13.21, -4.94]P+ 0.0%-7.82[-11.33, -4.54]P+ 0.0%-4.68[-9.95, +0.29]P+ 3.3%-4.63[-8.24, -1.37]P+ 0.1%-1.90[-4.83, +0.90]P+ 9.9%
28AnySat ❄21.00-24.05[-30.33, -17.68]P+ 0.0%-23.85[-30.39, -17.35]P+ 0.0%-23.67[-29.99, -17.52]P+ 0.0%-23.11[-28.82, -17.64]P+ 0.0%-22.81[-27.66, -18.04]P+ 0.0%-22.61[-27.49, -17.74]P+ 0.0%-20.08[-24.81, -15.53]P+ 0.0%-20.05[-24.59, -15.65]P+ 0.0%-19.95[-25.39, -14.37]P+ 0.0%-19.59[-25.38, -13.69]P+ 0.0%-19.42[-23.84, -15.03]P+ 0.0%-19.03[-24.36, -13.87]P+ 0.0%-18.90[-22.54, -15.54]P+ 0.0%-18.88[-25.48, -12.81]P+ 0.0%-17.36[-20.96, -14.05]P+ 0.0%-16.57[-20.42, -13.07]P+ 0.0%-15.67[-20.41, -11.61]P+ 0.0%-15.09[-18.96, -11.60]P+ 0.0%-14.99[-19.98, -10.47]P+ 0.0%-14.19[-18.05, -10.63]P+ 0.0%-13.23[-16.74, -9.82]P+ 0.0%-12.42[-16.82, -8.56]P+ 0.0%-11.32[-14.24, -8.58]P+ 0.0%-8.18[-13.17, -3.49]P+ 0.0%-8.13[-10.89, -5.64]P+ 0.0%-5.40[-8.67, -2.26]P+ 0.0%-3.50[-5.05, -1.82]P+ 0.0%
95% interval includes zero Row higher; pointwise interval excludes zero Row lower; pointwise interval excludes zero Mirrored half omitted

Method. 50,000 paired bootstrap replicates with random seed 20260908. In every replicate, the five training seeds are sampled with replacement and datasets are sampled with replacement within task type (4 classification, 4 segmentation, 2 change detection). The same draws are applied to every model. In-distribution averages two transfers, no-overlap averages three, superset averages two, and Overall gives equal weight to those three setting averages.

Interpretation. Highlighting is a pointwise ranking-stability diagnostic under resampling of the observed seeds and selected datasets. It is not a formal p-value and is not adjusted across the 378 comparisons. It does not include uncertainty from individual test samples.

Source: archived effective values from authoritative-sheet-batch-0.json through authoritative-sheet-batch-4.json, overlaid with the 810 uniquely keyed September 7 recovery evaluations in recovered-values.csv. The resulting means were checked against the current paired seed aggregates.