Share your thoughts, 1 month free Claude Pro on usSee more
WorkDL logo mark

Geo-Bench

Benchmarks

Task NameDataset NameSOTA ResultTrend
Generative Engine OptimizationGEO-Bench in-domain
Word Score25.42
30
Generative Engine OptimizationGEO-Bench
WLV4.81
26
Generative Engine OptimizationGEO-Bench Qwen-plus
Visual Fidelity Score10.17
13
Generative Engine OptimizationGEO-Bench Gemini-2.5-flash
Visibility15.35
13
Generative Engine OptimizationGEO-Bench GPT-4o-mini
Visual Fidelity18.31
13
Generative Engine OptimizationGEO-bench 1,000 queries (test)
Word Score11.07
12
Generative Engine OptimizationGEO-Bench Subjective Average (test)
VAR0.0116
12
Generative Engine OptimizationGEO-Bench Objective Overall (test)
VAR0.0189
12
Multispectral ClassificationGEO-Bench m-bigearthnet, m-so2sat, m-eurosat (test)
F1 Score (GB-ben)0.6215
10
Semantic SegmentationGEO-Bench SA-c
Macro mIoU28.98
10
Semantic SegmentationGeo-Bench
mIoU (nz-cattle, macro)82.98
10
Generative Engine OptimizationGEO-bench n=200 (test)
Citation Rate59.6
8
Subjective Content Quality AssessmentGEO-bench
Relevance Score22.3
2
Showing 13 of 13 rows