LangChain Changelog

Sign up for our newsletter to stay up to date

Powered by LaunchNotes

Loading...

April 2025

LangGraph v0.4: Working with Interrupts

Just shipped LangGraph v0.4! This release brings major upgrades for working with interrupts: • Interrupts are now surfaced automatically when you call...

LangSmith Self-Hosted

LangSmith Self-Hosted v0.10

This release brings alerting, UI‐driven experiment workflows, end-to-end OpenTelemetry support and a host of new capabilities alongside several bug fixes....

Catch Production Failures Early with Alerts in LangSmith

Set up real‑time notifications on error rates, run latency & feedback scores to spot failures before your customers do. If you’re already sending production...

LangChain Python updates: Improved content blocks, retry logic, and more

Now in LangChain Python: Better content blocks, retry logic, integrations & smarter agent support. Core updates: New integrations: Other improvements:

Run interactive evals with the LangSmith Playground

You can now create datasets inline and add examples to existing datasets without leaving the Playground. This update makes building a dataset to evaluate LLM...

March 2025

llms.txt Files and MCPDOC Server Launch for LangChain and LangGraph

We've introduced llms.txt files for LangChain and LangGraph, supporting both Python & JavaScript! These help your IDEs & LLMs access the latest...

Run evaluators in the LangSmith UI

We've made evaluations even easier to set up! You can now define evaluators for datasets and tracing projects directly in the UI —no code required. After...

Universal token counting callback for LangChain Python

We've added a new callback/context manager that tracks token usage across all major LangChain chat models, including support for: Docs:...

End-to-end native OpenTelemetry support for LangSmith SDK

Introducing End-to-End OpenTelemetry Support in LangSmith LangSmith now supports full end-to-end OpenTelemetry (OTel) for LangChain and LangGraph...

Structured output improvements for multiple models

A batch of improvements has been released for structured output integrations: The release covers three common methods for structured output:

Powered by LaunchNotes