LangChain

LangChain is the open-source framework and layered product ecosystem that helped define how developers build applications on top of large language models. Harrison Chase released the first version of LangChain, an 800-line Python package, from his personal GitHub account between October 16 and October 25, 2022, while working as an ML engineer at Robust Intelligence. The project caught fire as ChatGPT's release accelerated developer interest in LLM applications, and Chase incorporated LangChain as a company in January 2023 with co-founder Ankush Gola, quickly raising over $20 million from Sequoia Capital. The company has since grown into a three-layer product stack: LangChain (the original open-source framework for chains and RAG), LangGraph (a lower-level orchestration runtime for stateful, multi-agent systems, released in 2024), and LangSmith (an independent commercial observability and evaluation platform, launched in July 2023).

The core LangChain framework provides interoperable building blocks and the LangChain Expression Language (LCEL) for composing LLM calls, retrieval steps, and tool use into chains, with out-of-the-box support for parallelization, fallbacks, batching, and streaming. For more complex agentic workflows, LangGraph gives developers explicit control over state, branching, and multi-agent coordination, paired with LangGraph Studio, a visual IDE for building and stepping through an agent's decision graph. LangSmith sits above both, providing prompt-level tracing (Observability), systematic testing against datasets (Evaluation), and production hosting (Deployment), along with newer additions like LangSmith Fleet for managing many deployed agents and LangSmith Sandboxes. Both the LangChain and LangGraph frameworks are MIT-licensed and free to use, remaining valuable even with a single model provider by standardizing methods like fallbacks and async execution.

LangChain reports over 350 million monthly open-source downloads, more than 1 million developers building with its Python and JavaScript frameworks, and adoption by 7,000+ active LangSmith customers, including 5 of the Fortune 10. Companies such as Klarna, Rippling, Lyft, Gong, Harvey, Cisco, Workday, LinkedIn, and Coinbase use the ecosystem to build and observe production AI agents. The company raised a Series B in October 2025 at a $1.25 billion valuation, backed by Sequoia Capital and Benchmark among others, and hosts Interrupt, its annual agent-engineering conference. While the open-source frameworks remain free, LangSmith's commercial tiers monetize the observability and evaluation layer that most teams need to move agents reliably into production.

The LangChain and LangGraph frameworks are MIT-licensed, open source, and completely free to use with no usage limits from LangChain itself. The commercial layer, LangSmith, offers three tiers: Developer is free and includes 1 seat, 5,000 base traces/month, 14-day data retention, the prompt playground, and basic evaluations. Plus costs $39 per seat per month and includes 10,000 base traces included, full evaluations, custom dashboards, and email support; overage traces beyond the included allowance cost $2.50 per 1,000 (base, 14-day retention) or $5.00 per 1,000 (extended, 400-day retention), with base traces upgradeable to extended for an additional $2.50 per 1,000. Enterprise pricing is custom and adds dedicated support, custom retention policies, SSO, self-hosting, and higher trace volumes; exact figures require contacting LangChain sales. LangGraph Cloud/Platform deployment is billed separately from LangSmith, with plans starting around $35/month for hosted agent compute. Total production cost typically also includes underlying LLM API usage and vector database infrastructure, which are billed independently by those providers.