LangChain Sandbox: Run untrusted Python in your AI Agents
LangChain Sandbox lets you safely run untrusted Python in your AI agents! Built on Pyodide (Python in WebAssembly), LangChain Sandbox lets you execute code...
Node-level caching in LangGraph
You can now cache the results of tasks in LangGraph based on node input , helping you avoid redundant computation and speed up execution. How it works: See...
Integrate LangSmith prompts with your SDLC
You can already test, version, and collaborate on prompts in LangSmith. Now, you can automatically sync those prompts to GitHub, external databases, or CI/CD...
Monthly usage charts for easy tracking in LangSmith
We’ve rolled out monthly usage charts for SaaS customers in LangSmith. You can now track all billable metrics (as defined in Metronome) in one place on the...
Deferred nodes in LangGraph
LangGraph now supports deferred node execution , allowing you to create more sophisticated flows where a node only runs after all parallel branches have...
LangGraph Platform now supports MCP
Every deployed agent on LangGraph Platform now exposes its own MCP endpoint. Leverage your agents as tools in any client supporting streamable HTTP for MCP—...
LangGraph Platform GA: Deploy & manage long-running, stateful agents
LangGraph Platform is now generally available. Build, deploy, and manage long-running, stateful agents — at scale. LangGraph Platform is our purpose-built...
LangGraph Studio v2: Run and debug production traces locally
Our new and improved version of LangGraph Studio v2 allows you to: - Run locally (without using desktop app) - Edit prompts and other configuration directly...
Open Agent Platform: No-code platform to build agents
Open Agent Platform is our latest open-source, no-code agent building platform that lets you create, manage, and use LangGraph agents. Whether you're...
Agent Observability: Gain Insights into Tool Calls & Run Stats
We've launched new levels of visibility for agents to help you understand what your agent is doing! You can now get visibility into tool calls and run stats:...