LangSmith

What it does

LangSmith is presented by LangChain as a framework-agnostic observability and evaluation platform for LLM applications and AI agents. The product page describes tracing for preferred frameworks or any agent stack using Python, TypeScript, Go, or Java SDKs, with native tracing for popular agent frameworks and OpenTelemetry.

Product areas on the page include step-by-step agent tracing to pinpoint latency, cost, and quality issues, monitoring with cost tracking and online LLM-as-judge and code evals, tool and agent trajectory monitoring, and alerts via webhooks and PagerDuty. Insights features automatically cluster traces to detect usage patterns and failure modes, supported by SmithDB, a purpose-built database for nested agent traces.

Deployment and data notes on the page cover managed cloud hosted at smith.langchain.com with data stored in GCP us-central-1, plus bring-your-own-cloud and self-hosted options on Enterprise where data can run in a customer Kubernetes cluster on AWS, GCP, or Azure. The SDK uses an async callback handler so tracing does not block application performance.

Verified capabilities

Quick start

Trace an OpenAI call with the Python SDK

Install the langsmith package, enable tracing with environment variables from the SDK README, then wrap an OpenAI client so calls are traced automatically.

pip install -U langsmith

import openai
from langsmith.wrappers import wrap_openai

client = wrap_openai(openai.Client())
client.chat.completions.create(
    model="gpt-3.5-turbo",
    messages=[{"role": "user", "content": "Hello, world"}]
)

Source: https://pypi.org/project/langsmith/. Examples use placeholders only. Never paste a real key into a profile, config file you share, or a ticket.

MCP support: Current MCP support details were not cleanly verified for Braintrust in this research pass. Check the official Braintrust documentation for the supported setup before wiring an MCP client to it.

Works with

Only sourced support is listed. A missing framework means AgentsUse has not verified it yet, not that it cannot work.

Health and maintenance

Pricing and license

Freemium model with a free Developer tier, a Plus tier priced per seat per month with included base traces and pay-as-you-go usage beyond that, and custom Enterprise pricing. Usage is metered in LangChain storage and compute units tied to trace volume and retention, as described on the LangChain pricing page.

Official pricing or docs →

Limitations and safety

• Self-hosted and hybrid deployment are Enterprise-only according to the product FAQ and pricing summaries, while Developer and Plus are cloud-hosted.
• Pricing combines per-seat fees with per-trace usage and retention upgrades, so cost grows with both team size and trace volume.
• Deepest zero-config tracing is for LangChain and LangGraph stacks, while other frameworks rely on SDK wrappers or OpenTelemetry instrumentation.

Alternatives to LangSmith

Related tools

Common questions