Graphics Parsing on OmniParsingBench Graphics Module benchmark leaderboard · SOTA2 Research

Graphics Parsing on OmniParsingBench Graphics Module

87.43Overall AccuracyLogics-Parsing-Omni

Evaluation Results

Method	Links
Logics-Parsing-Omni2026.03		87.43	86.3	83.34	72.91	88.12	82.67	96.16	92.27	88.13	92.19
Logics-Parsing-Omni2026.03		87.43	—	—	—	—	82.67	—	—	—	92.19
Gemini-3-Pro2026.03		86.66	80.04	87.38	86.12	87.8	85.33	90.48	90.35	83.11	87.98
Gemini-3-Pro2026.03		86.66	—	—	—	—	85.33	—	—	—	87.98
GPT-5.22026.03		81.07	45.84	84.31	67.66	83.56	70.34	92.38	89.08	93.95	91.8
GPT-5.22026.03		81.07	—	—	—	—	70.34	—	—	—	91.8
Qwen3.5-397B-A17B2026.03		78.79	46.7	83.01	76.13	91.91	74.44	86.48	88.99	73.97	83.15
Qwen3.5-397B-A17B2026.03		78.79	—	—	—	—	74.44	—	—	—	83.15
Qwen3-VL-235B-A22B2026.03		78	52.82	84.17	67.62	83.99	72.15	86.69	83.12	81.74	83.85
Qwen3-VL-235B-A22B2026.03		78	—	—	—	—	72.15	—	—	—	83.85
Qwen3-Omni-30B-A3B2026.03		74.93	60.17	82.76	56.44	86.19	71.39	81.22	80.22	73.97	78.47
Qwen3-Omni-30B-A3B2026.03		74.93	—	—	—	—	71.39	—	—	—	78.47
Qwen3-VL-30B-A3B2026.03		73.05	62.34	82.1	53.41	67.4	66.31	82.67	82.26	74.43	79.78
Qwen3-VL-30B-A3B2026.03		73.05	—	—	—	—	66.31	—	—	—	79.78