LangChain
Open-source framework for building LLM applications
Overall score
About
LangChain is an open-source framework for developing applications powered by language models. It provides building blocks for chains, agents, retrieval-augmented generation, and tool use, plus a managed observability product (LangSmith) and a stateful agent runtime (LangGraph).
Best for: Engineering teams building LLM-powered applications who want a battle-tested framework for chains, agents, and RAG with strong observability and a path to production.
Pricing
Developer (OSS)
- Monthly
- n/a
- Annual /mo
- Free
- Billing
- flat
- Notes
- Full LangChain framework;LangGraph;5K free LangSmith traces/month;Community support · Open-source LangChain and LangGraph are free forever. LangSmith free tier covers individual builders.
Plus
- Monthly
- $39
- Annual /mo
- $39
- Billing
- per_seat
- Notes
- Higher LangSmith trace limits;Prompt management;Eval datasets;Email support
Enterprise
- Monthly
- n/a
- Annual /mo
- n/a
- Billing
- custom
- Notes
- Self-hosted LangSmith;SSO and SAML;Custom data residency;SOC 2 Type II;Dedicated support · Contact sales for pricing.
| Tier | Monthly | Annual /mo | Billing | Notes |
|---|---|---|---|---|
| Developer (OSS) | n/a | Free | flat | Full LangChain framework;LangGraph;5K free LangSmith traces/month;Community support · Open-source LangChain and LangGraph are free forever. LangSmith free tier covers individual builders. |
| Plus | $39 | $39 | per_seat | Higher LangSmith trace limits;Prompt management;Eval datasets;Email support |
| Enterprise | n/a | n/a | custom | Self-hosted LangSmith;SSO and SAML;Custom data residency;SOC 2 Type II;Dedicated support · Contact sales for pricing. |
Key features
- Composable chains and agents
- 200+ LLM and vector-store integrations
- LangGraph for stateful agent workflows
- LangSmith for tracing and evals
- Python and JavaScript SDKs
- Open-source core
Integrations
Trust & compliance
- Stage range
- n/a
- Founded
- 2022
- Status
- active
- SOC 2
- yes
- GDPR
- yes
- Data residency
- customer_choice
- External rating
- n/a
- Last verified
- Jun 2026
Reviews
Be the first to share your experience.
Related tools in Agent infrastructure
Pairs well with
People who've discussed LangChain
See all people →Curated mentions from podcasts, posts, and public stacks. Editorial coverage; not endorsements.
Andrew Ng
Founder; AI Educator
Andrew Ng partnered with LangChain founder Harrison Chase to build the AI Agents in LangGraph course and treats LangChain as part of the foundational toolkit for agentic AI. The LangChain ecosystem (LangChain + LangGraph + LangSmith) is the canonical reference architecture in Ng's Agentic AI curriculum.
Harrison Chase
Co-founder & CEO
Chase is co-founder and CEO of LangChain and the public voice for the agent framework. He shipped LangChain 1.0 in 2025 with revamped docs, general agent architectures and high-quality integrations, and continues to ship deep-research courses and DeepAgents Deploy as the no-code path to production agents. As CEO he's the structurally tightest endorsement surface for the framework.
Jason Zhou
Founder; Product Designer
Zhou has built an entire content pillar around LangChain on the AI Jason YouTube channel and his personal site, with a dedicated LangChain tutorial series covering how to build LLM applications and agent workflows. He repeatedly pairs LangChain with other primitives (Hugging Face for models, ElevenLabs for voice) in step-by-step tutorials that show working integrations. LangChain is the framework backbone of his AI experiments.
Martin Casado
General Partner
LangChain features in Casado's agent-infrastructure commentary. He frames LangChain as one of the canonical frameworks emerging on top of frontier models that lets enterprises move from prompt-engineering to actual agent-driven workflows in production.
Yohei Nakajima
General Partner
LangChain is referenced as part of the agent-orchestration stack in Nakajima's demos. He treats LangChain as a sibling primitive to BabyAGI for builders who want a more structured agent framework, with the choice between them often coming down to how much abstraction the builder wants over the raw LLM call pattern.