Project

gpt-researcher

assafelovic/gpt-researcher · Python · Apache-2.0

Point it at a question; it browses the web and hands back a sourced report.

GPT Researcher plans sub-questions, fires parallel web searches, scrapes and filters sources, then drafts a fully cited research report or markdown brief.

71.9 / 100 · Score
★ 30k Stars
+98 30-day growth
10 min Time to first run
Needs a paid API key API key

Score

Trend 55
Activity 81
Quality 95
Community 73
Relevance 55

Setup at a glance

DeployPip, Docker
GPUNo GPU needed
API keyNeeds a paid API key
UI languagesen
Doc languagesen
Install
pip install gpt-researcher

Deep dive

GPT Researcher is an autonomous research agent: give it a topic and it plans the research, scrapes and filters web sources, and produces a detailed, cited report — the kind of objective conclusion that used to take weeks of manual work.

It attacks the classic LLM failure modes directly: it grounds every claim in retrieved sources instead of hallucinating from outdated training data, aggregates many sources to get around token limits, and lets you pin which sources are allowed so bias stays under control.

It runs as a Python library, a CLI and an MCP server, and ships a Skill so agents in Claude Code and friends can invoke the same research loop; docs and README exist in six languages.

Use cases

Draft market or competitive research before a meeting
Summarise the state of the art on a niche topic
Generate a sourced brief from messy web info

Good for / Not ideal for

Good for

Analysts, founders and students doing deep dives
Teams that want citations, not vibes
Anyone tired of manually stitching search results

Not ideal for

Fully offline use (it needs web access)
Zero-cost setups (it calls an LLM + search API)

Trend · 30d