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

UI Agent Interaction on AgentNetBench

83.6Coordinate SR

Qwen-3-VL-72B

Updated 22d ago

Evaluation Results

Method	Links
Qwen-3-VL-72B 2026.05		83.6	68.3	51.9	80.2
Qwen-3-VL-72B 2026.05		82.5	67.4	50.8	79.3
Qwen3-VL-30B-A3B 2026.05		81.3	66	49.1	77.8
Qwen3-VL-30B-A3B 2026.05		80.4	65.1	48.2	77
Qwen-3-VL-72B 2026.05		79.2	64.5	48.1	76.9
ReVision 2026.05		77	61	44.2	73.8
Qwen3-VL-30B-A3B 2026.05		76.5	61.3	44.6	74.1
ReVision 2026.05		76.2	59.8	42.9	72.5
OpenCUA 2026.05		76.1	59.7	42.8	72.4
ReVision 2026.05		75.9	59.3	42.4	72
OpenCUA 2026.05		75.8	59.4	42.5	72.1
Qwen-3-VL-8B 2026.05		75.2	59.3	43.4	72.2
Qwen-3-VL-8B 2026.05		74.6	58.8	43.1	71.9
ReVision 2026.05		74.4	58.1	41.7	70.7
Qwen-3-VL-8B 2026.05		73.4	54.9	40.2	68.7
UI-TARS-1.5-7B 2026.05		72.3	54.5	40.6	68.9
OpenCUA 2026.05		72	55.6	39.8	68.4
UI-TARS-1.5-7B 2026.05		71.8	53.8	40.2	68.4
UI-TARS-1.5-7B 2026.05		70.4	52.3	38.8	66.9
UI-TARS-72B-DPO 2026.05		67.8	49.6	36	63.8
UI-TARS-72B-DPO 2026.05		67.1	49	35.5	63.2
UI-TARS-72B-DPO 2026.05		66.2	48.1	34.7	62.3
Kimi-VL-A3B 2026.05		55.2	40.4	28.1	50.9
Kimi-VL-A3B 2026.05		54	39.5	27.2	49.8
Kimi-VL-A3B 2026.05		51.6	37.8	25.9	47.2
Qwen2.5-VL-72B 2026.05		44.4	32.4	21.7	41.5
Qwen2.5-VL-72B 2026.05		43.6	31.8	21	40.8
Qwen2.5-VL-72B 2026.05		42.1	30.5	20.2	38.7
Qwen2.5-VL-32B 2026.05		41	30.5	20.2	38.7
Qwen2.5-VL-32B 2026.05		40.2	29.8	19.5	37.9
Qwen2.5-VL-32B 2026.05		38.5	28.2	18.4	35.6
Qwen2.5-VL-7B 2026.05		27	20.1	13.3	25.8
Qwen2.5-VL-7B 2026.05		26.4	19.5	12.7	25
Qwen2.5-VL-7B 2026.05		25.1	18.3	11.9	23.4