Gpt Researcher

Open source alternative to Perplexity AI, ChatGPT Deep Research and Gemini Deep Research

30.3K stars
Apache-2.0
Active this month
Last commit 18 days ago

Last verified Oct 29, 2026

Run autonomous deep research that searches web and local documents in parallel, synthesizes findings, and produces a cited report.

Visit website GitHub30.3K Deploy on Hostinger

Read the full breakdownWho it is for, how it works, trade-offs, the quick start and the FAQ

who it's for

Who Gpt Researcher is for#

Researchers and analysts building cited literature reviews

GPT Researcher queries academic databases including arXiv and PubMed alongside general web sources, then writes a multi-source report with inline citations. A single run on a focused topic produces a 2,000-word-plus document citing 20 or more sources, ready to export as a PDF or Word document.

Skip if:

Your literature review requires direct access to paywalled journal databases. GPT Researcher crawls publicly accessible pages; articles behind hard paywalls will not appear in results.

Developers integrating deep research into AI pipelines

The pip package and Python API allow GPT Researcher to run as a step in any async workflow. Pass a query, await conduct_research(), then call write_report() to get markdown output. The MCP server integration means any MCP-compatible host (Claude Desktop, Cursor, n8n) can trigger full research jobs from within an existing agent workflow.

Skip if:

Your pipeline needs sub-second responses. Each research run takes at minimum 45 seconds for filtering and report writing, plus retrieval time.

Teams researching internal document archives

Set DOC_PATH to a folder of PDFs, Word docs, Markdown files, or spreadsheets and run research queries against your own documents. Hybrid mode combines internal files with live web sources in a single report, which fits due diligence, competitive analysis, or workflows that need to reconcile internal data with current external information.

Skip if:

Your documents contain sensitive data that should not be sent to an external LLM API. Self-host with a local model via Ollama to keep all content on your infrastructure.

Claude and Cursor users who want richer research from MCP

The gptr-mcp server connects to Claude Desktop or Cursor in a few configuration steps, exposing deep_research, quick_search, and write_report as callable tools. Research initiated through the MCP server benefits from parallel retrieval and context filtering, producing more cited and thorough results than a plain web search tool.

Skip if:

You only need quick factual lookups. The overhead of a full GPT Researcher run is not worth it for simple one-line factual queries that a standard web search handles well.

the problem

The problem it solves#

Deep research at scale is slow and unreliable. Manual synthesis across many sources takes hours or days, and outputs reflect what you happened to find. LLMs trained on fixed snapshots hallucinate on recent or niche topics; their context windows are too short for multi-source reports, and commercial deep research services limit which sources they query, narrowing coverage and introducing selection bias.

Researchers and developers also lose control over the process. When queries run through a proprietary service, there is no visibility into which sources were scraped, which passages were cut, or why the model reached its conclusions. For sensitive topics, internal document research, or pipelines that must run on private infrastructure, those services are not viable options.

how Gpt Researcher solves it

How it solves it#

Parallel research architecture

A planner agent breaks any research question into sub-queries, then runs a separate crawler agent for each question in parallel. This reduces research time compared to sequential crawling and improves topic coverage by pursuing multiple angles simultaneously. The architecture scales with the MAX_ITERATIONS configuration.

20+ retriever integrations

Supported retrievers include Tavily, Google, Bing, Exa, arXiv, PubMed, DuckDuckGo, and SearXNG, as well as any MCP server. Multiple retrievers can run simultaneously in a single research job. Web search and local document research can be combined in one hybrid run.

Smart context filtering with Jev

Before any content reaches the LLM, GPT Researcher filters which passages are relevant to the query. The default filter, Jev by TypeSafe, kept 73% relevant passages versus 46% for embeddings across 28 benchmark tasks. When no Jev API key is set, it falls back to BM25 keyword ranking locally with no additional dependencies.

Local and hybrid document research

Research can run on local files instead of or alongside the web. Supported formats include PDF, plain text, CSV, Excel, Markdown, PowerPoint, and Word documents. Set the DOC_PATH environment variable and choose My Documents as the report source to search your own files.

Multiple export formats

Reports generate as markdown, PDF, Word documents (docx), JSON, and CSV. The average multi-agent run produces a 5-6 page report. For automated pipelines, the Python API returns the report as a string for downstream processing.

MCP server integration

GPT Researcher exposes a dedicated MCP server (gptr-mcp) that registers in Claude Desktop, Cursor, ChatGPT custom GPTs, and n8n. The server exposes tools for deep_research, quick_search, write_report, get_research_sources, and get_research_context to any MCP-compatible client.

strengths · trade-offs

Strengths and trade-offs#

Strengths

Apache-2.0 license with full self-hostingThe Apache-2.0 license allows commercial use, modification, and distribution without copyleft obligations. You can run GPT Researcher on your own servers, keep queries and documents private, and build products on top of it without license fees. Unlike Perplexity AI or OpenAI Deep Research, there is no per-query pricing or vendor dependency.
#1 on CMU DeepResearchGym benchmarkCarnegie Mellon University's DeepResearchGym benchmark evaluated leading deep research systems on 1,000 complex queries in May 2025. GPT Researcher scored highest in citation quality (precision 85.36%, recall 90.82%), report quality (clarity 83.70%, insightfulness 78.01%), and information coverage (key point recall 64.67%). It was the only system to exceed 85% on both citation precision and recall.
Any LLM provider, no code changes neededAny LLM provider that supports the OpenAI-compatible API format works as a drop-in. Supported providers include OpenAI, Anthropic, Google, Groq, Mistral, DeepSeek, and local models via Ollama, all configured through environment variables. No code changes are needed to swap providers.
Active community with academic citationsThe repository has 30,339 stars and 4,144 forks, with hundreds of contributors. The project has been cited in at least 10 academic papers published through July 2025, including papers from Carnegie Mellon's DeepResearchGym evaluation and the Johns Hopkins HLTCOE lab.

Trade-offs

-Requires LLM and retriever API keysTo run a research job, you need at least one LLM API key (such as OPENAI_API_KEY) and one retriever API key (such as TAVILY_API_KEY). A single deep research run costs approximately $0.40 using o3-mini on high reasoning effort. There is no fully free offline path; even local models still require a retriever unless you are only searching local files.
-Deep Research mode takes approximately 5 minutes per runThe recursive deep research mode explores topics in a tree, spawning sub-queries at each level. A single run takes approximately 5 minutes. This is expected behavior for thorough research, but it makes the tool unsuitable for workflows that need real-time or low-latency responses.
-Python 3.11+ environment requiredThe tool requires Python 3.11 or later. Teams running older Python environments need to upgrade or use the Docker deployment. The Docker Compose path simplifies this, but it introduces container overhead on resource-constrained servers.

versus alternatives

Gpt Researcher vs alternatives#

GPT Researcher vs Perplexity Deep Research

Both tools produce multi-source cited reports from web searches. GPT Researcher is open source (Apache-2.0) and runs on your own infrastructure; Perplexity Deep Research is a proprietary hosted service with no self-hosting option.

Feature	GPT Researcher	Perplexity Deep Research
License	Apache-2.0	Proprietary
Self-hosting	Yes	No
LLM choice	Any (OpenAI, Anthropic, local models)	Perplexity-selected
Retriever choice	20+ (Tavily, Google, arXiv, PubMed)	Perplexity-selected
Local document research	Yes	No
DeepResearchGym ranking (May 2025)	#1	Below GPT Researcher
Pricing	API key costs	Subscription

GPT Researcher is the better choice when you need source transparency, document privacy, control over which retrieval sources are searched, or integration into a custom pipeline. Perplexity Deep Research is worth considering when you want a polished hosted interface with no setup and are not concerned about where queries or sources are handled.

GPT Researcher vs ChatGPT Deep Research

Both tools use agentic research loops to produce long cited reports. The key differences are deployment model, LLM flexibility, and benchmark performance.

Feature	GPT Researcher	ChatGPT Deep Research
License	Apache-2.0	Proprietary
Self-hosting	Yes	No
LLM choice	Any provider	OpenAI models only
Local document research	Yes	Limited
MCP integration	Yes (gptr-mcp)	No
DeepResearchGym ranking (May 2025)	#1	Below GPT Researcher
Pricing	API key costs	ChatGPT Plus/Pro subscription

GPT Researcher is the better choice when you need to swap in a different LLM, connect research to a larger agent workflow via MCP, or keep your data off OpenAI's servers. ChatGPT Deep Research is the easier option for non-developers who want a chat-based interface with no local setup and are already paying for ChatGPT Plus.

install · quick start

Quick start#

bash

Install the pip package for library use, or run Docker Compose for the full server and frontend.
```bash
pip install gpt-researcher
docker compose up --build
```

tech stack · detected from GitHub

What it's built on#

Languages: JavaScriptPythonTypeScript
Frameworks: ExpressReact

frequently asked

FAQ#

Is GPT Researcher free to use?

Yes. GPT Researcher is Apache-2.0 licensed and free to self-host. The only operating cost is what you pay your LLM and retriever API providers. A single deep research run costs approximately $0.40 using o3-mini on high reasoning effort. You can reduce costs by using cheaper LLM providers or local models via Ollama.

How does GPT Researcher compare to Perplexity Deep Research and OpenAI Deep Research?

GPT Researcher ranked #1 on Carnegie Mellon University's DeepResearchGym benchmark (May 2025) across citation quality, report quality, and information coverage, ahead of both Perplexity and OpenAI Deep Research. Unlike those services, GPT Researcher is open source, runs on your own infrastructure, lets you choose any LLM and retriever, and does not charge per-query fees. A self-hosted deployment means your queries stay private and you can connect it to your own workflows.

What do I need to run GPT Researcher?

Python 3.11 or later, an LLM API key (such as OPENAI_API_KEY), and a retriever API key (such as TAVILY_API_KEY). You can use local models via Ollama for the LLM layer. Docker Compose is the easiest server setup if you do not want to manage a Python environment.

Can GPT Researcher research my own documents?

Yes. Set the DOC_PATH environment variable to a folder containing your files. Supported formats include PDF, plain text, CSV, Excel, Markdown, PowerPoint, and Word documents. You can run document-only research or combine local files with web sources in a single hybrid report.

Does GPT Researcher work with Claude or other AI assistants?

Yes. The official gptr-mcp repository provides an MCP server that registers with Claude Desktop, Cursor, ChatGPT custom GPTs, and n8n. Once configured, Claude can call deep_research, quick_search, and write_report directly from a conversation.

also worth a look

Similar open-source tools#

Vane

Alternative toPerplexity AI

AI-powered answering engine for private, cited web searches

37,793 GitHub stars
Licence MIT

hyperresearch

Alternative toChatGPT Pro

Deep research agent with a persistent vault for Claude Code

4,155 GitHub stars
Licence MIT

Local Deep Research

Alternative toPerplexity AI

Your AI research assistant, fully local and encrypted.

9,419 GitHub stars
Licence MIT

Scira

Alternative toPerplexity AI

Open source AI search engine that retrieves cited sources

11,907 GitHub stars
Licence AGPL-3.0

Hermes Agent

Featured

Alternative toGrok Bot

Self-hosted AI agent: persistent memory, multi-channel chat, model choice.

257,359 GitHub stars
Licence MIT

Happy

InDeveloper Tools

Open source AI coding orchestrator for Claude, Codex, Grok, Gemini.

24,386 GitHub stars
Licence MIT

by category

Open source tools in these categories#

423

Self-host Gpt Researcher on a Hostinger VPS

Keep it running 24/7, even when your laptop is off. A Hostinger VPS gives it a server of its own.

01
Choose a plan
Starts from $6.49/mo
02
Apply
Extra 10% off yearly plans
03
Install Gpt Researcher
On a server that stays on
04
Start building
Focus on the app, not the server

Deploy on Hostinger

Inbox Zero

Alternative toSuperhuman

AI email assistant that drafts replies and clears your inbox

12,826 GitHub stars
Licence AGPL-3.0
#Email

Omnara

Alternative toClaude Managed Agents

Open-source agent deployment API. Self-host or use Omnara Cloud.

3,296 GitHub stars
Licence Apache-2.0
#AI Agents

rakazo

Alternative toGrok Bot

AI teammates you own: your keys, your model, your machine.

5,914 GitHub stars
Licence Apache-2.0
#AI Agents

Advertise your product

The projects we add and every new guide. Free, unsubscribe any time.

Find open source alternatives to paid software

GPT Researcher vs Perplexity Deep Research

Feature	GPT Researcher	Perplexity Deep Research
License	Apache-2.0	Proprietary
Self-hosting	Yes	No
LLM choice	Any (OpenAI, Anthropic, local models)	Perplexity-selected
Retriever choice	20+ (Tavily, Google, arXiv, PubMed)	Perplexity-selected
Local document research	Yes	No
DeepResearchGym ranking (May 2025)	#1	Below GPT Researcher
Pricing	API key costs	Subscription

GPT Researcher vs ChatGPT Deep Research

Both tools use agentic research loops to produce long cited reports. The key differences are deployment model, LLM flexibility, and benchmark performance.

Feature	GPT Researcher	ChatGPT Deep Research
License	Apache-2.0	Proprietary
Self-hosting	Yes	No
LLM choice	Any provider	OpenAI models only
Local document research	Yes	Limited
MCP integration	Yes (gptr-mcp)	No
DeepResearchGym ranking (May 2025)	#1	Below GPT Researcher
Pricing	API key costs	ChatGPT Plus/Pro subscription

Who Gpt Researcher is for#

Researchers and analysts building cited literature reviews

Skip if:

Developers integrating deep research into AI pipelines

Skip if:

Teams researching internal document archives

Skip if:

Claude and Cursor users who want richer research from MCP

Skip if:

The problem it solves#

How it solves it#

Parallel research architecture

20+ retriever integrations

Smart context filtering with Jev

Local and hybrid document research

Multiple export formats

MCP server integration

Strengths and trade-offs#

Strengths

Trade-offs

Gpt Researcher vs alternatives#

GPT Researcher vs Perplexity Deep Research

GPT Researcher vs ChatGPT Deep Research

Quick start#

What it's built on#

FAQ#

Similar open-source tools#

Open source tools in these categories#

AI & Machine Learning

LLMOps & AI Tooling

Developer Tools

Find open source alternatives to paid software

Who Gpt Researcher is for#

Researchers and analysts building cited literature reviews

Skip if:

Developers integrating deep research into AI pipelines

Skip if:

Teams researching internal document archives

Skip if:

Claude and Cursor users who want richer research from MCP

Skip if:

The problem it solves#

How it solves it#

Parallel research architecture

20+ retriever integrations

Smart context filtering with Jev

Local and hybrid document research

Multiple export formats

MCP server integration

Strengths and trade-offs#

Strengths

Trade-offs

Gpt Researcher vs alternatives#

GPT Researcher vs Perplexity Deep Research

GPT Researcher vs ChatGPT Deep Research

Quick start#

What it's built on#

FAQ#

Similar open-source tools#

Open source tools in these categories#

AI & Machine Learning

LLMOps & AI Tooling

Developer Tools

Find open source alternatives to paid software