12k
Stars
1.7k
Forks
67
Open issues
26
Contributors
AI Analysis
Open Deep Research is an open-source deep research agent built with LangGraph that performs multi-step research workflows using configurable LLM providers and search tools. It serves specialized AI application developers and researchers who need to build or deploy autonomous research agents with comparable performance to commercial alternatives. This is not a general-purpose tool; it targets teams building agentic AI systems, not end-users seeking a simple search interface.
Inferred from signals mentioned in the README (tests, CI, type safety) — not a review of the actual code.
AI's overall editorial judgment — not an average of the bars above, can weigh other factors too.
LangChain's open-source deep research agent reaches PhD-level benchmark performance with flexible model/search configuration
Open Deep Research is a configurable, open-source autonomous research agent built on LangGraph. It accepts a research question, orchestrates multi-step web searches across configurable providers (Tavily, Bing, MCP servers, etc.), summarizes and compresses findings, and produces a structured report. It targets developers, AI researchers, and teams who want a self-hosted alternative to commercial deep research tools like Perplexity or OpenAI's Deep Research, with full control over model choice, search tooling, and pipeline logic. Maintained by LangChain, it sits at #6 on the Deep Research Bench leaderboard as of mid-2025.
Created November 2024 as deep-research agents emerged as a popular LLM application pattern. It evolved from simpler LangChain local-deep-researcher prototypes into a multi-model, multi-search pipeline with benchmark tracking added in 2025.
Initial growth was driven by LangChain's ecosystem reach and the viral 'deep research agent' trend in late 2024 and early 2025. A free course on LangChain Academy (August 2025), benchmark leaderboard placements, and GPT-5 integration updates sustained interest. Stars have plateaued relative to peak hype; 69 stars in 7 days suggests steady but not explosive ongoing growth.
Adoption not verified in production deployments at scale. The project is primarily positioned as a reference implementation and educational resource. LangChain Academy course and Deep Research Bench participation indicate community usage among practitioners, but no documented enterprise or high-volume production deployments are visible from README or metadata.
Appears to use a LangGraph state machine as its orchestration backbone, with separate LLM roles for summarization, research planning, web search execution, compression, and final report writing. Likely modular by design — configuration.py controls model and search provider selection. MCP compatibility suggests a plugin-style interface for search tools. Runs via a local LangGraph server with a Studio UI for inspection.
README references an evaluation script (tests/run_evaluate.py) and a LangSmith-based dataset of 100 PhD-level tasks for benchmarking, but unit/integration test coverage is not documented in README.
Last push was June 21, 2026, four days before evaluation date — actively maintained. The changelog in the README shows consistent feature updates through August 2025 (GPT-5, leaderboard entries, course launch). Repository appears healthy with no signs of stagnation.
ADOPT IF: you need a self-hosted, customizable research agent pipeline with model/search provider flexibility, LangGraph observability, and an active open-source community to learn from or contribute to. AVOID IF: you need production-grade reliability, SLA guarantees, or lack infrastructure to manage LLM API costs and server setup — commercial alternatives will be lower friction. MONITOR IF: you are evaluating whether open-source research agents can reach quality parity with commercial tools as benchmarks evolve and newer models are integrated.
Independent dimensions
Mainstream potential
4/10
Technical importance
7/10
Adoption evidence
3/10
- API cost unpredictability: running 100-question evaluations costs $20–$100; production use at scale may become expensive without careful query budgeting.
- Quality ceiling tied to underlying LLMs: benchmark rankings will shift as commercial model providers update their own offerings, potentially eroding the competitive positioning.
- LangChain ecosystem dependency: deep coupling to LangGraph means breaking changes in LangGraph or LangSmith could require significant adaptation effort.
- Niche positioning risk: as commercial deep research tools improve and lower barriers, the value proposition of self-hosting a comparable pipeline may narrow to privacy-sensitive or cost-sensitive use cases only.
- Sparse test coverage documentation raises uncertainty about regression handling when configurations or model providers change.
Likely to remain a well-maintained reference implementation and educational resource within the LangChain ecosystem. Mainstream dominance is unlikely given commercial competition, but it may become the de facto open-source benchmark baseline for Python-based research agents.
Newsletter
Get analyses like this every Monday
Free weekly digest of the most interesting open-source discoveries.
Languages
Information
- Language
- Python
- License
- MIT
- Last updated
- 2w ago
- Created
- 20mo ago
- Analyzed with
- anthropic/claude-haiku-4-5
Stars over time
Contributors over time
Top 100 contributors only — repos with more will plateau at 100.
Open issues
Top contributors
Recent releases
No releases published yet.
Similar repos
langchain-ai/deepagents
Deep Agents is an open-source agent framework that provides a production-ready...
Alibaba-NLP/DeepResearch
Tongyi DeepResearch is an open-source agentic LLM (30.5B parameters, 3.3B...
langchain-ai/local-deep-researcher
Local Deep Researcher is a fully local web research assistant that iteratively...
| Repository | Stars | Week Δ | Language | Score | Updated |
|---|---|---|---|---|---|
|
langchain-ai/open_deep_research
this
|
12k | +114 | Python | 8/10 | 2w ago |
|
langchain-ai/deepagents
|
26k | — | Python | 8/10 | 7h ago |
|
Alibaba-NLP/DeepResearch
|
19.6k | — | Python | 7/10 | 4mo ago |
|
langchain-ai/langgraph
|
37k | — | Python | 8/10 | 11h ago |
|
langchain-ai/local-deep-researcher
|
9.2k | — | Python | 8/10 | 2w ago |
 dzhng/deep-research
|
19.3k | — | TypeScript | 7/10 | 3mo ago |
A widely-starred TypeScript alternative with similar recursive search-and-summarize logic. Targets Node.js/web developers; open_deep_research targets Python/LangChain ecosystem users who want LangGraph observability and multi-model flexibility.
LangChain's own simpler sibling focused on fully local execution with Ollama. open_deep_research is the more capable, multi-provider variant with benchmark tracking; local-deep-researcher is the privacy-first, no-API-key option.
Higher star count, likely reflects stronger Chinese developer community traction and Alibaba model integration. open_deep_research has broader multi-provider flexibility and tighter LangSmith observability integration.
A broader LangChain agent framework with 25k stars. deepagents is a general-purpose agent toolkit; open_deep_research is a narrower, purpose-built research pipeline — they are complementary rather than directly competing.
Commercial offerings require no setup and may perform better on certain benchmarks. open_deep_research offers full transparency, customizability, and no per-query vendor lock-in — at the cost of infrastructure responsibility.