Grounding Model for UI Operations

AI that understands and interacts with user interfaces like humans do

Try It Now>
Three Actions, Infinite Possibilities

Three Actions, Infinite Possibilities

Click, Drag, and Scroll capabilities enable comprehensive control over any user interface, from simple buttons to complex workflows.

92.5%

Accuracy

92.5% on general tasks, 68.42% on challenging tasks

Accuracy

Tested on datasets: gboxai/cua-macos-benchmark

851ms

Speed

Ultra-fast response speed

Speed

Tested on datasets: gboxai/cua-macos-benchmark

$0.5/M

Cost

$0.50/M input and $0.50/M output

Cost

Playground

Evaluate Models for UI Operations

Select models (6)

GBOX
UITars
OpenAI CUA
Anthropic
Gemini
Gelato

Action

Click
Drag

The element you want to interact with