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

Geometric Reasoning on GeoAux-Bench-Core Olympiad Junior

95.17Accuracy

Qwen3-VL-235B-Thk.

Updated 1mo ago

Evaluation Results

Method	Links
Qwen3-VL-235B-Thk. 2026.03		95.17
Qwen3-VL-30B-Thk. 2026.03		93.24
Qwen3-VL-235B-Ins. 2026.03		91.79
Qwen3-VL-30B-Ins. 2026.03		89.37
GPT-5 2026.03		88.41
Gemini-2.5-Pro 2026.03		88.24
Doubao-seed-1.6 2026.03		86.47
GLM-4.5V 2026.03		82.61
Gemini-2.5-Flash 2026.03		82.13
MiMo-VL-7B-SFT 2026.03		80.68
MiMo-VL-7B-RL 2026.03		80.68
InternVL3.5-8B 2026.03		66.67
Claude-Sonnet-4.5 2026.03		60.39
Claude-Opus-4.1 2026.03		58.45
InternVL3-8B 2026.03		55.56
GLM-4.1V-Thk. 2026.03		53.14
Qwen2.5-VL-Ins. 2026.03		52.17
MathCanvas-7B 2026.03		51.69
GPT-4o 2026.03		42.51
BAGEL-7B-MoT 2026.03		38.16
BAGEL-Zebra-CoT 2026.03		35.27