My 8 ChatGPT Agent tests produced only 1 near-perfect result - and a lot of alternative facts
Briefly

OpenAI introduced the ChatGPT Agent, enhanced from Deep Research and Operator capabilities, available at $200/month for Pro members, with a limit of 400 interactions. Plus users can access it at $20/month with 40 interactions. Initial tests showed that Agent generally understood queries, yet often struggled to deliver valuable results. Notably, it fails in managing large-scale projects, particularly in data analysis tasks, which limits its usability despite a few instances of exceptional output.
The new ChatGPT Agent combines capabilities of Deep Research and Operator, allowing interaction with user interface elements beyond previous versions.
ChatGPT Agent costs $200/month for Pro tier users, providing 400 agent interactions, whereas Plus users will have access to 40 interactions.
Testing showed that while Agent comprehended questions, it often produced inadequate results, except in one instance where the output was impressively useful.
Agent struggles with larger projects and complex tasks, particularly in data analysis, indicating limitations in its current capabilities.
Read at ZDNET
[
|
]