OpenAI Launches ChatGPT Agent: Capable of Proactive Thinking and Tool Selection, Paving the Way for a New Era of Autonomous Task Processing

On July 18, OpenAI unveiled the ChatGPT Agent - a groundbreaking advancement that empowers ChatGPT with autonomous reasoning and action capabilities, bridging research insights with practical applications.

By integrating the strengths of prior Operator modules and deep research tools, the ChatGPT Agent operates within a virtual computing environment to manage complex workflows. It autonomously selects appropriate tools from its comprehensive toolkit to handle financial analysis, form completions, presentation creation, and seamlessly transitions between reasoning phases and execution modes. Users can now request competitive landscape analysis with professional-grade presentations or real-time calendar updates for client meetings, all with editable, polished outputs.

User control remains central to the operation. The agent requests authorization before executing critical actions, allowing users to pause tasks, take manual control of browsers, or terminate operations at any time. Equipped with diverse web tools, it dynamically selects optimal execution paths while maintaining contextual awareness across different toolsets through continuous process optimization.

Performance benchmarks demonstrate impressive capabilities: the Excel-compatible .xlsx format support achieved 45.54% overall score on SpreadsheetBench - comparable to Microsoft Excel Copilot. It established industry records in WebArena and BrowseComp benchmarks and outperformed previous models in investment banking analyst tasks.

Practical applications span both daily and professional scenarios, automating repetitive tasks like financial statement updates and travel planning. Pro, Plus, and Team subscribers gain immediate access while Enterprise and Education users receive authorization in July. Pro users enjoy nearly unlimited task executions monthly, with other paid plans allowing 50 automated tasks per month.

Security features include mandatory user confirmation, active monitoring, and risk-mitigation protocols to prevent adversarial attacks and misuse. OpenAI emphasizes this launch as the foundation for continuous evolution, with planned iterations to enhance the ChatGPT Agent's utility through iterative improvements.