Remote Sensing Perception and Reasoning on XLRS-Bench benchmark leaderboard · SOTA2 Research

Remote Sensing Perception and Reasoning on XLRS-Bench

53.1Average Score (Avg.)GeoLLaVA-8K+DualComp

Evaluation Results

Method	Links
GeoLLaVA-8K+DualComp2026.04		53.1	26.7	45	37.4	53	69.5	43.6	34	65	69	79	72	48.3	49
Intern-S1-mini2026.04		51.6	—	—	—	—	—	—	—	—	—	—	—	—	—
GeoLLaVA-8K2026.04		51.5	26.7	38	35	49	69	41.6	31.6	65	67	78	66	50	52
Qwen2.5-VL-72B2026.04		50.2	33.3	47	34	39	80	45.3	42.1	65	71	74	37	43.3	42
Qwen3-VL-8B2026.04		50	21.7	50	30.4	26	81.5	46.6	43.1	66.7	74	79	37	43.3	51
GPT-5.22026.04		47.5	30	37	34	17	70.5	43	41.4	68.3	74	76	52	36.7	38
Qwen2.5-VL-7B2026.04		47.4	33.3	40	36.2	31	77	40.6	40.5	66.7	68	72	27	38.3	45
GeoLLaVA-8K+FastV2026.04		47.1	23.3	41	37.4	46	50.5	43.8	31.6	65	53	60	65	46.7	49
InternVL2.5-8B2026.04		46.2	38.3	37	21.6	10	77	33.4	35.5	65	73	83	34	50	43
Qwen2-VL-7B2026.04		45.8	26.7	40	31.8	11	73	35.9	34.6	61.7	70	81	35	46.7	48
InternVL3-8B2026.04		45.6	40	39	25.2	10	71.5	44.5	30.8	65	77	82	36	21.7	50
Gemini 2.5 Pro2026.04		45.2	—	—	—	—	—	—	—	—	—	—	—	—	—
GeoLLaVA-8K+VisionZip2026.04		42.8	23.3	39	38.6	37	49	44.1	30	65	36	38	62	46.7	47
LLaVA-Next+RFM+DIP2026.04		42.2	36.7	41	25.6	2	54.5	33.9	34	53.3	70	76	24	53.3	44
LLaVA-Next2026.04		41.7	26.7	40	30	5	67	28.8	32.8	66.7	69	78	27	35	36
GeoLLaVA-8K+SparseVLM2026.04		41.2	21.7	39	38.8	38	32.5	33.8	26.5	65	43	43	62	45	47
Claude 3.7 Sonnet2026.04		40.5	27.6	22.7	27.6	17.4	68.4	30.5	29.9	63.6	64.8	78.4	34.5	27.8	32.6
ZoomEarth2026.04		40.2	—	—	—	—	—	—	—	—	—	—	—	—	—
GeoChat2026.04		22.9	16.7	29	24.2	2	23	21.1	16.8	35	33	43	10	—	21