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

Geometry problem solving on Geo3K

98.2Top-1 Accuracy

Draw2Think BL

Updated 13d ago

Evaluation Results

Method	Links
Draw2Think BL 2026.05		98.2
Draw2Think CT 2026.05		95.3
Human expert 2026.05		90.9
OpenAI-o3 2026.05		81
Gemini-2.5-Pro 2026.05		80.7
GPT-5-Thinking 2025.09		79.9
Pi-GPS 2026.05		77.8
Gemini-2.5-Pro 2025.09		77.2
PGPSNet-v2 2026.05		76.4
GeoDRL 2026.05		68.4
GeoGen-SFT 2026.05		58.4
Gemini 2.0 Flash 2026.02		58.2
InterGPS 2026.05		57.5
GeoFocus-7B 2026.02		55.3
VAPO-Thinker-7B 2025.09		51.3
VLAA-Thinker-7B 2025.09		50.6
GeoFocus-3B 2026.02		50.4
GeoUni 2026.02		50
Vision-R1-7B 2025.09		49
GeoGen-SFT-7B 2026.02		46.3
R1-OneVision-7B 2025.09		46.1
InternVL2.5-8B 2025.09		44.8
Qwen2.5-VL-7B 2026.05		44.8
VAPO-Thinker-3B 2025.09		44.2
Qwen2.5-VL-7B 2026.02		39.4
GeoGen-SFT-3B 2026.02		38.6
Qwen2.5-VL-7B 2025.09		38.5
G-LLaVA-13B 2026.02		35
GPT-4o 2026.02		34.6
InternVL-3-8B 2026.02		33.9
GPT-4o 2026.05		31.5
Claude 3.7 2026.02		31.1
InternVL-3-2B 2026.02		31.1
Gemma-3-4B 2026.02		30.6
GeoSketch-RL 2026.05		28.8
Qwen2.5-VL-3B 2026.02		25.3
InternVL-2.5-8B 2026.02		16.6
mPLUG-Owl2-7B 2026.02		12.7
Deepseek-VL-7B 2026.02		5.2
Monkey-Chat-7B 2026.02		4.7
Chameleon-7B 2026.02		3.9