Hallucination Evaluation on MVH-Bench Cross-Instance benchmark leaderboard · SOTA2 Research

Hallucination Evaluation on MVH-Bench Cross-Instance

73.66Binary AccGPT-4o

Evaluation Results

Method	Links
GPT-4o2026.03		73.66	52.41	32.04	57.27	36.11	251.49
InternVL3-14B2026.03		64.68	40.69	18.33	68.47	47.59	239.76
Gemini 2.5 Flash2026.03		64.42	39.68	14.91	60.83	38.15	217.99
InternVL3-8B2026.03		64.21	38.43	16.76	69.58	48.43	237.41
Qwen2.5-VL-7B2026.03		64.1	41.3	16.29	66.57	43.52	231.78
GPT-4o mini2026.03		62.2	31.53	12.87	66.99	44.81	218.4
Claude 4.5 Sonnet2026.03		61.44	36.94	13.61	60	36.02	208.01
Qwen2-VL-7B2026.03		61.07	35.18	12.87	61.99	37.87	208.98
Gemini 2 Flash2026.03		60.72	29.58	9.63	68.15	44.72	212.8
LLaVA-OneVision-7B2026.03		58.36	32.36	10.93	62.78	36.29	200.72
Deepseek-VL-Chat-7B2026.03		58.17	29.3	7.22	60.69	27.96	183.35
LLaVA-NeXT-I-7B2026.03		56.85	30.05	8.34	60.14	33.05	188.43
Gemma3-12B2026.03		55.76	18.52	3.89	59.21	33.24	170.62
Mantis-8B-Idefics22026.03		55.21	28.29	7.96	56.94	28.98	177.38
Llama-3.2-11B-V-I2026.03		54.6	20.74	4.81	59.72	33.52	173.39