List your company

GUI Grounding on ScreenSpot Pro

72.7AccuracyGemini-3-Pro
47.11653.75860.467.042May 11, 2025Jul 19, 2025Sep 26, 2025Dec 4, 2025Feb 11, 2026Apr 21, 2026Jun 29, 2026

Evaluation results

MethodLinks
Gemini-3-ProAccess=API call only2026.0372.7————
vLLMBase Model=GUI-Owl-1.5-32B-Instruct2026.0570.97———3.3
AQuaUIBase Model=GUI-Owl-1.5-32B-Instruct2026.0569.89———2.77
UI-Venus-1.5-30B-A3Bdesign=end-to-end2026.0269.6————
vLLMBase Model=GUI-Owl-1.5-8B-Instruct2026.0568.88———2.69
UI-Venus-1.5-8Bdesign=end-to-end2026.0268.4————
UI-Venus-1.5-8BAccess=Open weights2026.0368.4————
quality-aware self-distillation2026.0668.37————
MAI-UI-32Bdesign=end-to-end2026.0267.9————
AQuaUIBase Model=GUI-Owl-1.5-8B-Instruct2026.0567.68———2.21
Naive-OPSD2026.0666.41————
ZoomOnceSize=4B2026.0666.2————
Holo2-30B-A3Bdesign=end-to-end2026.0266.1————
MAI-UI-8Bdesign=end-to-end2026.0265.8————
MAI-UI-8BAccess=Open weights2026.0365.8————
SFT2026.0665.15————
MAI-UISize=4B2026.0665.1————
GUI-SD2026.0664.71————
vLLMBase Model=MAI-UI-8B2026.0564.52———2.15
Seed1.8design=end-to-end2026.0264.3————
GTA1-32Bdesign=end-to-end2026.0263.6————
GRPO2026.0663.5————
EvoCUA-OpenCUA-72BBase Model=OpenCUA-72B2026.0163.24————
Qwen3.5-9BParameters=9B2026.0663————
Step-GUI-8Bdesign=end-to-end2026.0262.6————
Step-GUI-8BParameters=8B2025.1262.6————
UI-Venus-1.0-72Bdesign=end-to-end2026.0261.9————
UI-TARS 1.52025.0561.6————
UI-TARS-1.5Version=1.52025.1261.6————
UI-TARS 1.5-7BAccess=Open weights2026.0361.6————
Vero Q3I-8BRL=Qwen3VL, Initial Model=8B Inst2026.0461.4————
MolmoPoint-GUI-8BGUI-specific Fine-tuning=true, Test-time scaling=64crops2026.0361.1————
Seed 1.5-VL2025.0560.9————
SeedVL-1.5Version=1.52025.1260.9————
OpenCUA-72BParameters=72B2026.0160.8————
OpenCUA-72Bdesign=end-to-end2026.0260.8————
GUI-SDTime/epoch=4.22026.0560.7————
Qwen3-30B-A3B-InstructParameters=30B, Arch=A3B, Type=Instruct2025.1260.5————
MolmoPoint-GUI-8BGUI-specific Fine-tuning=true2026.0360.2————
Step-GUI-4Bdesign=end-to-end2026.0260————
Step-GUI-4BParameters=4B2025.1260————
Qwen3-VL-4B-InstructParameters=4B, Type=Instruct2025.1259.5————
AQuaUIBase Model=MAI-UI-8B2026.0559.14———1.88
Holo2-8Bdesign=end-to-end2026.0258.9————
Holo2-8BAccess=Open weights2026.0358.9————
vLLMBase Model=MAI-UI-2B2026.0558.89———1.25
Propose-then-Critic-8B2026.0558.7————
vLLMBase Model=Qwen3-VL-32B-Instruct2026.0558.57———3.24
Image-LoRABackbone=QWEN2.5-VL-72B, Text:Img Token Ratio=1:5, Avg #Image Tokens=4,195, FLOPs (G)=75.82025.1258.35————
Std-LoRABackbone=QWEN2.5-VL-72B, Text:Img Token Ratio=1:5, Avg #Image Tokens=4,195, FLOPs (G)=496.92025.1258.18————
GUI-Owl-32Bdesign=end-to-end2026.0258————
GUI-Owl-32BParameters=32B2025.1258————
GUI-OwlSize=32B2026.0658————
UI-Venus-1.5-2Bdesign=end-to-end2026.0257.7————
AQuaUIBase Model=MAI-UI-2B2026.0557.5———1.25
MAI-UI-2Bdesign=end-to-end2026.0257.4————
Qwen3-VL-Instruct + GRPO-GaussianTime/epoch=16.82026.0557.4————
Qwen3-VL-32B-ThinkingParameters=32B, Thinking Mode=true2026.0157.1————
UI-InsSize=7B2026.0657————
ZwZ-8B2026.0556.8————
Qwen3-VL-Instruct + GRPO-BinaryTime/epoch=16.92026.0556.8————
Qwen3-VL-Instruct + GRPO-DistanceTime/epoch=16.72026.0556.6————
vLLMBase Model=Qwen3-VL-8B-Instruct2026.0556.23———1.95
AQuaUIBase Model=Qwen3-VL-32B-Instruct2026.0556.17———2.44
Image-LoRABackbone=QWEN2.5-VL-72B, Text:Img Token Ratio=1:4, Avg #Image Tokens=3,367, FLOPs (G)=59.492025.1255.94————
OpenCUA-32Bdesign=end-to-end2026.0255.3————
OpenCUASize=32B2026.0655.3————
AQuaUIBase Model=Qwen3-VL-8B-Instruct2026.0555.28———1.67
GUI-Owl-7Bdesign=end-to-end2026.0254.9————
GUI-Owl-7BParameters=7B2025.1254.9————
GUI-OwlSize=7B2026.0654.9————
Qwen3-VLSize=32B2026.0654.9————
Qwen3-VL-8B-InstructParameters=8B, Type=Instruct2025.1254.6————
Std-LoRABackbone=QWEN2.5-VL-72B, Text:Img Token Ratio=1:4, Avg #Image Tokens=3,367, FLOPs (G)=415.52025.1254.04————
Qwen3-VL-30B-A3Breproduction=true, design=end-to-end2026.0253.7————
Qwen3-VL-Instruct2026.0553.6————
Ferret-UI LiteSize=–2026.0653.3————
ZoomUI-7B2026.0552.8————
Qwen3-VL-8Breproduction=true, design=end-to-end2026.0252.7————
Qwen3-VL-8BAccess=Open weights2026.0352.7————
Molmo2-GUI-8BAccess=Fully open, GUI-specific Fine-tuning=true2026.0352.3————
UI-Ins-7B2026.0552.2————
InfiGUI-G1Size=7B2026.0651.9————
UI-Venus-1.0-7Bdesign=end-to-end2026.0250.8————
UI-VenusSize=7B2026.0650.8————
Std-LoRABackbone=QWEN2.5-VL-72B, Text:Img Token Ratio=1:3, Avg #Image Tokens=2,534, FLOPs (G)=333.72025.1250.6————
JEDI-7BAccess=Fully open2026.0350.2————
GroundCUA-7BAccess=Fully open2026.0350.2————
GTA1-7Bdesign=end-to-end2026.0250.1————
GTA1-7BAccess=Fully open2026.0350.1————
GTA1-7B2026.0550.1————
GTA1Size=7B2026.0650.1————
Image-LoRABackbone=QWEN2.5-VL-72B, Text:Img Token Ratio=1:3, Avg #Image Tokens=2,534, FLOPs (G)=43.82025.1250.09————
OpenCUA-7Bdesign=end-to-end2026.0250————
OpenCUA-7BAccess=Fully open2026.0350————
OpenCUASize=7B2026.0650————
EvoCUA-32BParameters=32B2026.0149.76————
Qwen3-VLSize=8B2026.0649.5————
HyperClick-7B2026.0548.2————
GPT-4o + OS-Atlas-7BType=Planner+Grounder, Params=≫7B2026.0348.1————