Search-based Visual Question Answering on WebEyes 1.0 (test) benchmark leaderboard · SOTA2 Research

Search-based Visual Question Answering on WebEyes 1.0 (test)

83.33Accuracy (Vehicles)Gemini-3.1-Pro

Evaluation Results

Method	Links
Gemini-3.1-Pro2026.05		83.33	48.05	44.3	68	62.4	79.22	63.82
Doubao-Seed-1.82026.05		69.44	28.57	41.77	78.8	54.4	64.94	61.34
Doubao-Seed-2.0-Pro2026.05		69.44	42.86	45.57	81.2	56.8	68.83	65.37
Doubao-Seed-1.62026.05		63.89	29.87	41.77	66.8	53.6	53.25	54.97
GPT-4.12026.05		58.33	24.68	37.97	40	39.2	51.95	40.22
GPT-4o2026.05		58.33	24.68	27.85	20.4	39.2	40.26	29.97
Gemini-3.1-Flash-Lite2026.05		55.56	23.38	43.04	40.4	40.8	49.35	40.68
Qwen3-VL-8B2026.05		47.22	24.68	31.65	35.6	43.2	38.96	36.34
GPT-5.42026.05		44.44	28.57	35.44	60	48.8	53.25	49.38
SophiaVL-R12026.05		38.89	23.68	22.78	27.46	34.4	37.66	29.67
Pixel-Searcher2026.05		38.89	24.68	34.18	50.4	42.4	42.86	42.24
OneThinker-8B2026.05		36.11	21.05	29.11	24.59	40	23.38	28.26
VL-Rethinker2026.05		33.33	23.68	31.65	27.05	33.6	29.87	29.2
UniVG-R12026.05		30.56	19.74	31.65	22.13	36	22.08	26.22
InternVL-3.5-8B2026.05		25	29.87	32.91	42	36	31.17	36.02
Vision-R12026.05		13.89	22.37	17.72	27.87	12.8	19.48	21.19