Tactile Property Prediction and Commonsense Reasoning on VTV-150K benchmark leaderboard · SOTA2 Research

Tactile Property Prediction and Commonsense Reasoning on VTV-150K

81.2HardnessTouchThinker

Evaluation Results

Method	Links
TouchThinker2026.06		81.2	77.1	78.2	61.5	46.4	80.5	68.3	52.9	80	69.5
TouchThinker2026.06		79.1	80.8	75.3	63.4	40.7	78.9	64.2	50.7	74	67.4
TacReasoner2026.07		78.16	77.82	76.54	58.46	40.35	73.56	70.32	54.27	71	66.7
TacReasoner2026.07		77.84	78.62	79.76	54.75	41.58	76.93	74.89	58.41	75	68.6
VTV-LLM2026.06		73.9	75	67.3	56.5	35.6	71.3	57.6	43.2	64	60.4
VTV-LLM2026.07		73.9	75	67.3	56.5	35.6	71.3	57.6	43.2	64	60.4
VTV-LLM2026.06		72.1	78.2	68.1	52.8	38.2	72.1	59.7	45.9	72	62.1
VTV-LLM2026.07		72.1	78.2	68.1	52.8	38.2	72.1	59.7	45.9	72	62.1
Gemini-2.5-Pro-Exp2026.06		36.2	34.7	39.1	21	4.3	42.6	29.4	18.5	40	29.5
Gemini-2.5-Pro-Exp2026.07		36.2	34.7	39.1	21	4.3	42.6	29.4	18.5	40	29.5
GPT-52026.06		34.7	32.6	32.6	18.7	2.1	40.9	38.4	16.6	36	28
GPT-4o2026.07		34.7	32.6	32.6	18.7	2.1	40.9	38.4	16.6	36	28
LLaVA-Video-Qwen22026.06		30.4	29.7	28.9	18.1	2.1	33.6	29.4	17.2	36	25
LLaVA-Video-Qwen22026.07		30.4	29.7	28.9	18.1	2.1	33.6	29.4	17.2	36	25
LLaVA-OneVision2026.06		27.5	32.6	26	20.2	0.7	40.9	28.2	11.7	30	24.2
LLaVA-OneVision2026.07		27.5	32.6	26	20.2	0.7	40.9	28.2	11.7	30	24.2
Qwen2.5-VL2026.06		25.3	28.9	17.3	15.9	1.4	22.9	28.2	16	30	20.6
Qwen2.5-VL2026.07		25.3	28.9	17.3	15.9	1.4	22.9	28.2	16	30	20.6
InternVL2.5-VL2026.06		18.1	23.9	21	13.7	0	24.5	17.9	11.1	24	17.1
InternVL2.5-VL2026.07		18.1	23.9	21	13.7	0	24.5	17.9	11.1	24	17.1
VideoLLaMA32026.06		15.2	21.7	14.4	10.8	0	11.4	12.8	7.4	20	12.6
VideoLLaMA32026.07		15.2	21.7	14.4	10.8	0	11.4	12.8	7.4	20	12.6