List your company

GUI Agent Interaction on OSWorld

42.5Average AccuracyUI-TARS
7.45216.55125.6534.749May 16, 2025Jul 20, 2025Sep 23, 2025Nov 28, 2025Feb 1, 2026Apr 7, 2026Jun 12, 2026

Evaluation results

MethodLinks
UI-TARSModel=UI-TARS-1.5-72B, Steps=100, Open Source=Close source2025.0542.5——————————
OpenAI CUAModel=GPT-4o, Steps=200, Open Source=Close source2025.0538.1——————————
UI-TARSModel=UI-TARS-1.5-72B, Steps=50, Open Source=Close source2025.0538——————————
InfantAgent-NextModel=Claude-3.7-Sonnet + UI-TARS-1.5-7B, Steps=50, Open Source=Open source2025.0535.3——————————
Agent S2Model=Claude-3.7-Sonnet, Steps=50, Open Source=Open source2025.0534.5——————————
OpenAI CUAModel=GPT-4o, Steps=50, Open Source=Close source2025.0532.6——————————
DynamicUIBackbone=Qwen3-vl-8B, Step=502026.0428.434.869.61719.626.113.162.54017.743.5
Opencua-7BStep=502026.0428.238.643.613.232.633.312.143.542.228.347.1
DynamicUIBackbone=UITARS-1.5-7B, Step=502026.0428.232.857.78.539.341.38.629.85027.554.4
Claude Computer UseModel=Claude-3.7-Sonnet, Steps=100, Open Source=Close source2025.0528——————————
UITARS-1.5-7BStep=502026.0427.232.853.98.539.341.38.625.646.724.452.2
AGMem (Ours)Backbone Model=GPT-5.4-mini2026.0627.2——————————
UI-TARSModel=UI-TARS-1.5 7B, Steps=100, Open Source=Open source2025.0526.9——————————
Claude Computer UseModel=Claude-3.7-Sonnet, Steps=50, Open Source=Close source2025.0526——————————
UITARS-72b-dpoStep=502026.0425.833.261.512.825.543.56.733.333.323.547.8
Qwen3-vl-8B*Step=502026.0425.823.934.81731.856.59.233.366.740.421.7
UI-TARSModel=UI-TARS-72B-DPO, Steps=50, Open Source=Open source2025.0524.6——————————
+ visual memoryBackbone Model=GPT-5.4-mini2026.0620.4——————————
Opencua-a3bStep=502026.0419.921.653.92.123.334.86.537.56.711.843.5
Qwen3-vl-4B*Step=502026.0419.830.421.712.821.839.15.233.346.72313
Vanilla agentBackbone Model=GPT-5.4-mini2026.0618.3——————————
o3Step=502026.0417.221.738.58.54.321.711.837.52017.621.7
AGUVISModel=AGUVIS 72B + GPT-4o, Open Source=Open source2025.0517——————————
Aria-UIModel=Aria-UI + GPT-4o, Open Source=Open source2025.0515.1——————————
OS-AtlasModel=OS-Atlas-Base-7B + GPT-4o, Open Source=Open source2025.0514.6——————————
SeeClickModel=SeeClick + GPT-4o, Open Source=Open source2025.059.21——————————
Qwen2.5Model=Qwen2.5-vl-72B, Open Source=Open source2025.058.8——————————