Visual Question Answering on UniPercept-Bench-VQA ISTA 1.0 (test) benchmark leaderboard · SOTA2 Research

Visual Question Answering on UniPercept-Bench-VQA ISTA 1.0 (test)

89.74Scene AccuracyUniPercept

Evaluation Results

Method	Links
UniPercept2025.12		89.74	85.71	82.44	93.94	78.51	82.69	89.24	78.54	83.12	85.51	84.23
GLM-4.5-V2025.12		81.2	79.67	74.55	72.73	75.21	80.77	76.74	73.68	79.87	78.26	77.22
InternVL32025.12		79.06	85.16	77.42	69.7	78.51	81.41	79.65	73.68	84.42	81.16	79.28
LLaVA-OneVision-1.5-Instruct2025.12		78.63	85.16	82.44	72.73	80.17	83.33	81.4	75.3	84.42	88.41	81.13
Claude-Sonnet-4.52025.12		77.35	78.02	73.12	87.88	75.21	76.28	74.71	74.09	81.82	76.81	76.08
Claude-Sonnet-4.52025.12		76.92	78.57	74.91	90.91	77.69	76.92	77.03	74.49	81.82	79.71	77.32
Gemini-2.5-pro2025.12		76.5	82.42	77.06	66.67	77.69	78.21	78.2	75.71	82.47	71.01	77.73
GPT-4o2025.12		75.64	79.12	73.48	33.33	77.27	71.79	78.78	69.23	77.92	72.46	74.64
InternVL32025.12		75.64	79.12	73.48	33.33	77.27	71.79	78.78	69.23	77.92	72.46	74.64
QwenVL-2.5-Instruct2025.12		74.79	72.53	74.91	51.52	73.55	73.72	77.33	66.8	74.03	73.91	73.3
Llama-4-Scout2025.12		73.5	75.27	71.68	72.73	67.77	75.64	69.77	69.64	77.27	69.57	71.86
InternVL3.52025.12		54.27	50.55	58.42	39.39	36.36	46.79	56.69	48.58	29.87	71.01	49.79
InternVL3.52025.12		50	55.49	61.29	30.3	35.95	50.64	59.3	42.91	37.01	57.97	50.1
QwenVL-3-Instruct2025.12		27.78	32.42	25.45	39.39	24.79	14.74	23.26	28.34	25.32	81.16	27.63
Random Guess2025.12		26.5	23.63	24.73	30.3	30.58	26.28	23.84	24.29	33.77	33.33	26.6
QwenVL-3-Instruct2025.12		26.5	24.73	19	15.15	18.6	11.54	18.31	22.67	17.53	66.67	21.65
QwenVL-2.5-Instruct2025.12		14.1	29.12	19.71	12.12	18.6	20.51	12.21	14.57	31.17	46.38	19.59