Categories Companies Research

Loading the SOTA2 catalog…

Explore

Categories Companies Research

Contribute

Join SOTA2 Contact

SOTA2

© 2026 SOTA2

Build in Bellevue, USA

WizWand is now SOTA2 Research

Papers Benchmarks Datasets Tasks

Geometric Reasoning research benchmarks · SOTA2 Research

Geometric Reasoning

Benchmarks

Dataset NameSOTA MethodSort

Dataset NameSOTA methodMetricTrendResultsLast Updated

VL-Cogito-7B + LEAD

69.1Accuracy@1

42

Apr 3, 2026

Geometry3K (test)

40.93Accuracy

22

Mar 25, 2026

GeoAux-Bench Core (Total)

83.16Accuracy

21

Mar 20, 2026

GeoAux-Bench-Core Olympiad Junior

Qwen3-VL-235B-Thk.

95.17Accuracy

21

Mar 20, 2026

GeoAux-Bench-Core Olympiad Senior

Qwen3-VL-235B-Thk.

89.45Accuracy

21

Mar 20, 2026

GeoAux-Bench-Core Curriculum Junior

84.75Accuracy (%)

21

Mar 20, 2026

GeoAux-Bench-Core Curriculum Senior

83.91Accuracy

21

Mar 20, 2026

NoisyRollout-7B

55.2Accuracy

16

Jun 23, 2026

GeoSketch-Qwen2.5VL-7B

72.52Accuracy

15

May 12, 2026

28.9Accuracy@1

15

Feb 26, 2026

34.6Overall Accuracy

13

Mar 13, 2026

45Accuracy

12

Mar 24, 2026

44.11Avg@8 Accuracy

10

Apr 15, 2026

Geometry3K Geo-Sub-Aux (test)

52.4Accuracy

8

Jun 25, 2026

Geometry3K Geo-Sub (test)

50.9Accuracy

8

Jun 25, 2026

BLINK Rel. Depth

ROBOALIGN (Ours)

87.9Score

7

Mar 24, 2026

OlympiadBench-Geo olympiad level

77.68Accuracy

6

Mar 27, 2026

Geometry3K mid-school level

81.03Accuracy

6

Mar 27, 2026

GeoQA mid-school level

79.58Accuracy (GeoQA mid-school)

6

Mar 27, 2026

47Score

5

Mar 23, 2026

Geometric Shapes

84.5Final Accuracy

4

Jun 24, 2026

GeoSym-Bench (test)

Qwen3-VL-8B + GeoSym

18.79Accuracy (%)

4

May 19, 2026

Qwen3-VL-7B-Instruct (GeoCode trained)

26.49Accuracy

4

Feb 26, 2026

MathVerse plane-geometry

Qwen3-VL-7B-Instruct (GeoCode trained)

65.1Accuracy

4

Feb 26, 2026

BBH Geometric Shapes

Zero-shot (Default Imp.)

53Accuracy

3

Jun 2, 2026

Popular datasets

Geometry3K GeoAux-Bench GeoQA Geomverse BBH

Follow for updates

Join SOTA2 Contact