Vision-Language Reasoning and Knowledge on ConBench benchmark leaderboard · SOTA2 Research

Vision-Language Reasoning and Knowledge on ConBench

76.51AccuracyQwen3.0-VL-8B

Evaluation Results

Method	Links
Qwen3.0-VL-8B2026.05		76.51	1	21.41	33.45	31.17
InternVL-3.5-8B2026.05		74.53	2	18.83	30.06	31.24
Qwen2.5-VL-7B2026.05		73.54	3	39.15	51.1	47.03
InternVL-2.5-8B2026.05		72.05	4	21.8	33.48	27.86
Qwen2.0-VL-7B2026.05		71.75	5	16.55	26.9	23.37
InternVL-3.0-8B2026.05		71.46	6	44.2	54.62	51.01
Gemma3.0-12B2026.05		70.76	7	46.18	55.89	53.91
InternVL-2.0-8B2026.05		64.52	8	28.05	39.1	30.82
Qwen-VL-7B-Chat2026.05		56.89	9	28.84	38.28	28.07
LLaVA-1.5-13B2026.05		53.02	10	13.38	21.37	13.12
LLaVA-1.6-13B2026.05		46.78	11	27.25	34.44	22.33