Share your thoughts, 1 month free Claude Pro on us
See more
Home
/
Benchmarks
Geometry Problem Solving on UniGeo calculation (test)
Loading...
96.9
Top-1 Accuracy
Draw2Think CT
40.116
54.858
69.6
84.342
May 20, 2026
Top-1 Accuracy
Updated 13d ago
Evaluation Results
Method
Method
Links
Top-1 Accuracy
Draw2Think CT
Backbone=gemini-3-flas...
2026.05
96.9
Draw2Think BL
Backbone=gemini-3-flas...
2026.05
93.6
Gemini-2.5-Pro
Protocol=Zero-shot, Re...
2026.05
79.6
OpenAI-o3
Protocol=Zero-shot, Re...
2026.05
79.3
GeoGen-SFT
Backbone=qwen2.5-vl-7b...
2026.05
78
GeoSketch-RL
Backbone=qwen2.5-vl-7b...
2026.05
72.5
NeSyGeo
Backbone=qwen2.5-vl-7b...
2026.05
71.8
GeoDRL
Protocol=Neural-symbol...
2026.05
66.7
Qwen2.5-VL-7B
Protocol=Zero-shot, Re...
2026.05
64.2
SocraticGeo
Backbone=qwen2.5-vl-7b...
2026.05
49.2
GPT-4o
Protocol=Zero-shot, Re...
2026.05
42.3
Feedback
Search any
task
Search any
task