Omnidirectional Image Understanding on ODI-Bench open-ended 1.0 (test) benchmark leaderboard · SOTA2 Research

Omnidirectional Image Understanding on ODI-Bench open-ended 1.0 (test)

49.53Overall Scoreo3

Evaluation Results

Method	Links
o32025.10		49.53	55.49	45.36	69.5	50	52.89	45.4	34.59	39.6	62.04	59.1
GPT-4o2025.10		42.91	52.62	39.74	68.5	45.35	43.8	32.27	27.25	30.3	61.84	49.5
InternVL3-78B2025.10		42.52	47.05	43.91	62	57.56	48.76	35.58	29.77	29.82	53.27	53.95
Intern-s12025.10		42.12	46.3	41.68	53.5	63.95	40.5	34.72	31.87	30.8	52.04	58.85
InternVL3-38B2025.10		40.96	48.11	39.57	59	40.94	38.43	34.66	29.77	27.6	54.08	52.62
Qwen-VL-Plus2025.10		39.87	44.39	39.35	67.5	47	36.36	34.39	29.1	27.65	51.09	46.2
Qwen2.5-VL-72B2025.10		39.49	45.89	35.14	63	42.44	39.55	34.91	24.1	27.6	54.39	47.75
Qwen2.5-VL-3B2025.10		38.91	40.8	39.64	63.5	41.86	41.32	35.77	28.09	26.8	47.35	56.2
Qwen2.5-VL-32B2025.10		37.67	39.56	26.34	62	42.44	30.91	36.47	30.21	30.43	50.41	44.25
Gemini-2.0-flash2025.10		36.42	37.82	28.55	48.26	50	56.5	30.86	25.89	31	55.51	42.17
glm-4v-9B2025.10		35.79	38.8	26.12	60.5	42.44	32.23	36.38	25.26	26.5	41.43	42.95
LLava-Next-34B2025.10		35.25	41.57	29.28	49.5	41.52	7.025	35.4	20.21	24.6	45.51	52.42
MiniCPM-V 4.02025.10		32.52	36.02	30.56	57	43.6	38.43	29.34	24.32	21.04	46.53	36.25
LLaVA-v1.5-7B2025.10		32.29	35.02	32.07	56.5	24.42	9.504	28.28	25.47	26.6	45.1	43.5
InternVL2.5-8B2025.10		30.86	33.52	22.34	62	39.53	34.71	21.96	25.68	20.4	38.98	51.55
LLava-Next-7B2025.10		30.52	36.58	32.24	60.5	40.12	12.81	21.04	22.01	17.6	38.78	44.45
XComposer22025.10		28.36	30.12	20.92	48	34.88	1.652	23.13	25.79	23	42.65	43.15
idefics3-8B2025.10		27.93	31.5	28.16	65	43.6	27.27	16.75	15.41	20.6	37.35	37.9
Deepseek-VL-1.3B2025.10		27.8	31.33	18.22	53.5	33.14	6.198	20.98	19.92	23.1	42.24	44.15
LLaVA-ov-0.5B2025.10		17.9	28.75	11.12	42	19.19	3.719	6.196	5.765	8.3	29.18	32.2