Share your thoughts, 1 month free Claude Pro on usSee more

Spatial and Temporal Reasoning on MMSI-Bench (test)

43Cam-Cam Accuracy

LLaVA-OneVision-72B

Updated 5d ago

Evaluation Results

Method
LLaVA-OneVision-72B 2026.04	43	31.9	33.3	30.2	37.6	38.6	28.1	19.7	13.5	32.9	15.7	28.4
InternVL3-78B 2026.04	34.4	23.4	32.1	12.8	37.6	26.5	37.5	19.7	28.4	31.6	29.3	28.5
InternVL2.5-8B 2026.04	32.3	27.7	29.6	32.6	24.7	32.5	26.6	27.3	16.2	31.6	30.3	28.7
InternVL2.5-4B 2026.04	31.2	23.4	21	31.4	34.1	25.3	23.4	24.2	13.5	31.6	26.8	26.3
DeepSeek-VL2-Tiny 2026.04	29	27.7	21	23.3	17.6	31.3	14.1	24.2	14.9	25	27.3	24
InternVL2.5-2B 2026.04	28	27.7	24.7	37.2	29.4	36.1	43.8	15.2	21.6	31.6	26.8	29
STCR-Tag 2026.04	27.9	33	32.1	33.7	36.5	32.1	29.7	24.2	16.2	34.2	27.3	29.6
Qwen2.5-VL-3B 2026.04	26.9	27.7	30.9	29.1	28.2	34.9	31.2	16.7	17.6	27.6	23.2	26.5
InternVL3-2B 2026.04	26.9	25.5	29.6	31.4	28.2	27.7	26.6	22.7	12.2	23.7	23.7	25.3
Llama-3.2-11B-Vision 2026.04	25.8	30.8	32	25.6	21.2	25.9	20.3	19.7	25.6	28.9	19.2	25.4
InternVL3-8B 2026.04	25.8	31.9	37	25.6	35.3	28.9	23.4	24.2	16.2	32.9	14.6	25.7
STCR-CoT 2026.04	25	30.2	24.7	25.3	34.1	30.2	28.6	25.8	20.3	35.5	26.3	28.3
Qwen2.5-VL-32B 2026.04	24.7	26.6	29.6	22.1	32.9	31.3	31.2	24.2	18.9	35.5	27.8	27.7
InternVL2.5-26B 2026.04	24.7	19.1	29.6	33.7	31.8	37.3	35.9	30.3	10.8	31.6	26.8	28
Qwen2.5-VL-7B 2026.04	24.7	24.5	24.7	25.6	29.4	26.5	25	18.2	20.3	39.5	25.8	25.9
InternVL3-1B 2026.04	24.7	35.1	22.2	30.2	29.4	30.1	32.8	28.8	17.6	19.7	26.3	27
DeepSeek-VL2-Small 2026.04	24.7	28.7	18.5	33.7	38.8	27.7	28.1	33.3	24.3	25	29.8	28.6
InternVL2.5-78B 2026.04	23.7	22.3	39.5	29.1	31.8	42.2	35.9	19.7	17.6	26.3	27.3	28.5
InternVL2.5-1B 2026.04	23.7	26.6	24.7	25.6	31.8	25.3	31.2	30.3	17.6	25	26.3	26.1
DeepSeek-VL2 2026.04	23.7	31.9	22.2	36	30.6	22.9	28.1	15.2	28.4	26.3	28.3	27.1
LLaVA-OneVision-7B 2026.04	20.4	33	29.6	29.1	25.9	30.1	29.7	25.8	18.9	34.2	11.6	24.5
InternVL3-14B 2026.04	19.4	24.5	24.7	23.3	37.6	24.1	31.2	22.7	24.3	31.6	29.3	26.8
InternVL2.5-38B 2026.04	18.3	22.3	35.8	22.1	38.8	34.9	37.5	25.8	14.9	38.2	25.3	27.9
InternVL3-9B 2026.04	18.3	25.5	32.1	29.1	31.8	22.9	29.7	24.2	16.2	38.2	26.8	26.7
NVILA-8B 2026.04	17.2	29.8	24.7	30.2	22.4	34.9	34.4	25.8	25.7	34.2	29.8	28.1