tools
LangChain Review 2026 — Top Open-Source LLM Framework
LangChain is the most popular open-source framework for building LLM apps. 90K+ GitHub stars, 600+ integrations, LangGraph for agents, LangSmith for observability.
In inglese
LangChain provides pre-built agent architectures, integrations, middleware, and a durable runtime for building applications that use language models, tools, and data sources.
Developers use it to create agents, connect models and tools, add human approval, persist state, and customize agent behavior. LangChain itself is free and MIT-licensed; LangSmith services for tracing, evaluation, deployment, and other operations are priced separately.
Features
- Pre-built agent architectures and templates
- Integrations with models, tools, databases, and data sources
- Middleware hooks for approval, context compression, and data removal
- Persistence, rewind, checkpointing, and human-in-the-loop support
- Model and tool swapping without rewriting applications
- MIT-licensed open-source library
- Integration with LangSmith for tracing, evaluation, and deployment
Use cases
- Build agents that call tools and model providers
- Connect agents to databases and external data sources
- Add human approval to sensitive agent actions
- Persist and resume long-running agent workflows
- Route requests between models based on task requirements
Pros
Cons
Latest updates
- The completed items list in an annotation queue now offers Add to Dataset for selected thread items, importing each thread as a multi-turn example.
The completed items list in an annotation queue now offers Add to Dataset for selected thread items.
- Edit and delete in the experiments table now check project permission, to match BE.
Edit and delete in the experiments table now check project permission.
- Charts now show shape-aware loading skeletons instead of temporary zero values while data loads. Evaluator spend charts use bar and metric placeholders that preserve the final layout.
Charts now show loading skeletons instead of temporary zero values while data loads.
- Downloading experiment results with “Include annotator name and comment per row” now keeps automated evaluator scores in every feedback column, alongside human annotations.
Downloading experiment results now keeps automated evaluator scores in every feedback column, alongside human annotations.
- The evaluators list API accepts an agent_id filter to return evaluators attached to the agent’s environments or tagged datasets.
The evaluators list API accepts an agent_id filter.
Capabilities
- Command line — “Sandbox CLI” source
- Choice of models — “Choose your model.” source
- Self-hosted — “Self-hosted and hybrid deployment options” source
- API — “Trigger agent via API” source
- Builds agents and workflows — “Build agents fast with or without code. Choose your model. Own your harness.” source
- Traces and evaluates — “Debug, monitor, and improve your AI applications with tracing and evals” source
Get it
Security
Pricing
- Starting price
- $39/user/mo
- Prices checked
- 2026-09-24
Developer
Free
- Up to 5k base traces / mo, then pay-as-you-go
- Community support
- 1 seat
Plus
- $39 / seat per month
- Up to 10k base traces / mo, then pay-as-you-go
- Access to Deployment, Engine, and more
- Add unlimited seats
- 1 free Serverless (Small) deployment included
Enterprise
Price on request
- Self-hosted and hybrid deployment options
- Custom SSO, ABAC, and RBAC
- Support SLA
- Custom seats and workspaces