What it does
Gives the model the skill set of a research assistant: it pulls from Google Scholar, PubMed, arXiv, JSTOR and others (200M+ resources), helps you find papers from a rough description, break down a PDF, extract the methodology and assemble a reference list with citations. Its edge is breadth of sources — think of it as a science search engine layered on top of ChatGPT.
In the store it has 2M+ chats, making it one of the most popular research GPTs.
How to set it up
- Open chatgpt.com/g/g-kZ0eYXlJe-scholar-gpt → Start Chat.
- Some advanced actions (analysing large PDFs, exporting) run through the author's premium features; basic search works right away.
When to use it
- ✅ Literature reviews, tracking down a primary source from a vague description, "explain this paper in plain language" walkthroughs.
- ✅ When you need coverage across many academic databases at once, not just one.
- ❌ Not a replacement for Consensus when you want a synthesis of the evidence on a question — Scholar is stronger at finding and unpacking specific papers.
Example
"Find the key 2023-2025 papers on LLM agents in product analytics and turn them into an APA reference list" — it returns a set of papers with links and formats the bibliography.