ChatGPT add New 'Deep Research' Feature

OpenAI has just add new feature into ChatGPT called "deep research," this tool can independently dive into complex topics and emerge with well-researched answers – complete with citations and methodology explanations.

You can feed it questions, images, PDFs, or spreadsheets, and it will spend anywhere from 5 to 30 minutes crafting a detailed response. The process is transparent too – you can watch its research journey unfold in a sidebar as it works.

To prove its worth, the company reports that their deep research model achieved a remarkable 26.6% accuracy on "Humanity's Last Exam" – a benchmark for expert-level questions. That's a massive leap from GPT-4's previous 3.3% score.

But this power comes at a premium. The feature is primarily aimed at Pro users ($200/month), who get up to 100 queries monthly. Plus, Team, and Enterprise users will have limited access, with expanded access promised once a faster, more cost-effective version is developed.

OpenAI is honest about the tool's limitations – it may occasionally "hallucinate" facts or struggle to distinguish between authoritative information and rumors.

However, this launch, following their recent "Operator" browser tool, signals a clear push toward making AI more practical and valuable for everyday professional use.