LangChain
MIT-licensed framework for building LLM agents in Python and JavaScript
Overview
LangChain is an MIT-licensed framework for building LLM applications and agents in Python and JavaScript, with more than 1,000 integrations for models, tools and data stores. Since the 1.0 release in October 2025, its agents run on the LangGraph runtime; LangChain Inc. sells LangSmith for tracing, evaluation and deployment on top.
Pricing
Pricing shown for reference only. These figures reflect RECATOOLS research as of 19 Oct 2026 and may be out of date or incomplete. This is not financial or purchasing advice — always confirm the current price on the provider’s official website before making any decision.
- LangChain (Python and JavaScript)
- LangGraph durable runtime
- deepagents for long-running agents
- Up to 5k base traces a month, then pay as you go
- Community support
- Up to 10k base traces a month, then pay as you go
- 1 free Serverless (Small) deployment; access to Deployment and Engine
- Add unlimited seats
- Self-hosted and hybrid deployment
- Custom SSO, ABAC and RBAC
- Support SLA, custom seats and workspaces
- 0.005 LSU per base trace (14-day retention)
- 0.0025 LSU to upgrade a trace to 180-day retention
- Deployment, sandbox and Fleet usage metered in LSUs
Use cases
What you can produce with LangChain
- Agents built with create_agent on any supported model provider, swappable without rewriting the app
- Connections to more than 1,000 model, tool, vector-store and data integrations
- Durable agent state with checkpointing, rewind and human-in-the-loop approval through LangGraph
- Middleware that summarises long conversations, redacts sensitive data or adds approval steps
- Retrieval-augmented generation over your own documents
- Traces and evaluations of the same agents in LangSmith, switched on with one environment variable
- Long-running autonomous agents through the deepagents framework
LangChain is the quickest way to get an agent working across many model providers, and since 1.0 it is a steadier dependency: one create_agent entry point, LangGraph's durable runtime underneath, frequent releases and a public promise of no breaking changes before 2.0. LangSmith adds tracing and evals with a single environment variable.
The framework is free under MIT; the bill comes from LangSmith seats and usage. For simple applications a provider SDK may be leaner, and teams that want control usually work in LangGraph directly. A sensible default for prototyping agents, with a clear path into production.
What we found
LangChain is the open-source framework for building LLM applications and agents in Python and JavaScript. Harrison Chase released it in October 2022, and LangChain Inc. now maintains it from San Francisco. It wraps model providers, tools, vector stores and retrievers behind common interfaces, and its product page counts more than 1,000 integrations. The 1.0 releases of LangChain and LangGraph on 22 October 2025 changed its shape. Agents now go through a single create_agent function running on LangGraph's durable runtime, with persistence, checkpointing and human-in-the-loop interrupts, plus middleware for jobs such as summarising long conversations or redacting sensitive data. Older chains moved to a separate langchain-classic package, and the team committed to no breaking changes before 2.0.
That promise has held so far. On 19 October 2026 the Python package stood at 1.4.6, langchain-core at 1.6.12 and LangGraph at 1.2.16, each released within the previous week and all MIT-licensed. The company now presents three open-source frameworks side by side: LangChain for getting an agent running quickly, LangGraph for low-level control, and deepagents for long-running autonomous work.
The framework is free; the business is LangSmith, a platform for tracing, evaluation, deployment, sandboxes, an LLM gateway and no-code agents. Developer is free for one seat with 5,000 base traces a month, Plus is $39 per seat a month with 10,000, and Enterprise is custom and the only route to self-hosting. Usage beyond that is billed in LangChain Standard Units at $1 each, with a base trace costing 0.005 units. LangChain raised $125 million at a $1.25 billion valuation in October 2025, and its homepage claims 350 million monthly downloads across its frameworks.
The long-standing criticism is abstraction: for a simple app, a provider SDK means less code and easier debugging, and moving chains into a legacy package concedes part of that. For agents that need state and resumability, LangGraph is the stronger argument.
Compiled by RECATOOLS from public sources: vendor documentation and pricing, code repositories and published reporting.
Notable facts
- LangChain reached 50,000 GitHub stars within 3 months of release, making it one of the fastest-growing developer frameworks in history.
- Harrison Chase built the first version of LangChain in a weekend as an experiment and open-sourced it without expecting widespread adoption.
- LangChain's npm package is downloaded over 1 million times per week, making it one of the most-used AI libraries in the JavaScript ecosystem.
Frequently asked questions
About this listing
This entry was compiled from publicly available data including LangChain's official website, press releases, documentation, and reputable third-party publications. RECATOOLS is not affiliated with LangChain unless explicitly stated.
Third-party AI tools update their pricing, features, availability, and policies frequently. Information here may be outdated by the time you read this — we make reasonable efforts to keep listings current, but cannot guarantee absolute accuracy.
For the latest details, please refer to LangChain directly →
Spotted something out of date? Suggest an update →
Related reading
More in Agents & Automation
