Geometric diagram parsing on SGDP-1K (test) benchmark leaderboard · SOTA2 Research

Geometric diagram parsing on SGDP-1K (test)

99.4Point PrecisionGemini-3-Flash

Evaluation Results

Method	Links
Gemini-3-Flash2026.04		99.4	99.7	99.6	96.3	96	96.1	88.7	70.3	78.5	91.9	92.8	92.4	88.9	91.1
Qwen3-VL-235B-Thinking2026.04		99.2	95.2	97.1	86.7	75.3	80.6	33.3	77.2	46.6	94.5	45.1	61	83.2	73.7
GDP-4B-RL2026.04		99.2	99.3	99.2	96.1	96.8	96.4	86.7	86.2	86.5	95.8	95.2	95.5	97	94.9
Qwen3-VL-4B-Instruct2026.04		99	98.5	98.8	48.7	69.5	57.3	44.7	47.2	46	35	73.9	47.5	70.8	64
Qwen3-VL-32B-Instruct2026.04		99	99.1	99	63.6	77.9	70	42.9	91.1	58.4	27.5	82.1	41.2	83.8	70.4
Qwen3-VL-8B-Instruct2026.04		98.9	99.1	99	42.3	65.4	51.4	48.9	57.6	52.9	32	75.1	44.8	77.5	65.1
GPT-5.2-12112026.04		98.9	98.5	98.7	78.2	68.1	72.8	57	76.2	65.3	81.2	71.3	75.9	72.7	77.1
GDP-4B-SFT2026.04		98.9	99.2	99	95.7	95.9	95.8	83.3	84.2	83.7	93.5	94.5	94	96.6	93.8
Qwen3-VL-235B-Instruct2026.04		98.7	93.9	96.2	75.3	73.2	74.2	51.5	83.1	63.6	72.6	75.5	74	83	78.2
InternVL3.5-8B-Instruct2026.04		98.5	94.8	96.6	57.8	50.4	53.8	23.3	86.1	36.7	62.4	72.6	67.1	69.7	64.8
Claude-4.5-Sonnet2026.04		95.7	96.1	95.9	70.1	72.6	71.8	28.5	85.2	42.7	64.8	73.4	68.8	78.3	71.5
LLaVA-OneVision-1.5-7B2026.04		91.6	96.9	94.2	54.8	66.5	60.8	34.5	57.4	43.1	31.2	62.3	41.5	69.6	61.9