OpenAI has introduced Operator, an AI agent that excels at performing tasks within web browsers thanks to its innovative Computer-Using Agent (CUA) model based on GPT-4o. Operator leverages advanced vision capabilities to understand visual content and interact with user interfaces. It operates through a cycle of perception, reasoning, and action while ensuring user safety by requiring direct intervention for sensitive tasks like password entry. Despite its record performance on several benchmarks, Operator still does not match human capabilities in these tasks, highlighting the ongoing gap between AI and human proficiency.
OpenAI's Operator achieves new state-of-the-art performance using a new model, CUA, allowing the AI agent to interact with web browsers effectively.
With Operator, users will have an AI agent capable of performing complex web tasks autonomously while maintaining necessary safety protocols and user control.
Collection
[
|
...
]