LangSmith
LangChainObservability and testing platform from the LangChain team, providing tracing, datasets, evaluations, a prompt hub and annotation queues, alongside deployment and sandbox infrastructure for agents. It integrates tightly with the LangChain ecosystem and instruments any application through OpenTelemetry SDKs, and is offered as managed cloud, a customer-managed cloud (BYOC) option, or a self-hosted enterprise deployment.
Openness
1 high confidence- license
- Proprietary
- source
- closed
- platform
- proprietary(no public source for LangSmith itself)
LangSmith the observability/eval platform is closed/proprietary; only the LangChain/LangGraph frameworks (separate registry entries) are open. Self-host/BYOC are licensed deployments of closed software, not source-available.
- https://docs.langchain.com/langsmith/self-hosted recorded 2026-09-29
Self-hosted docs page: "Self-hosted LangSmith is an add-on to the Enterprise plan designed for our largest, most security-conscious customers" and it mentions a license key to trial it. Searched the body for open source LangSmith or a public repository for LangSmith itself: the only open source mention is PostgreSQL as a dependency.
Adoption
4 medium confidenceLangSmith is the default observability platform for the LangChain ecosystem, and the vendor names an enterprise roster including Klarna, Lyft and Coinbase, but publishes no user, customer or download count. The langsmith PyPI package is excluded because it is a hard dependency of langchain-core, so its downloads measure LangChain installs rather than LangSmith use.
- https://www.langchain.com/langsmith recorded 2026-10-08
The customer logo strip still carries Klarna, Lyft and Coinbase among others (logo images and alt text), positioned as the LangChain-ecosystem observability platform. No user, customer or download count is stated.
Capability
4 medium confidenceLangSmith offers framework-native tracing across Python, TypeScript, Go and Java, monitoring with cost tracking and LLM-as-judge evaluations and alerts, prompt and dataset workflows, and deployment, fleet and sandbox tooling for agents, keeping it level with Langfuse.
- https://docs.langchain.com/langsmith/evaluation recorded 2026-09-29
Evaluation docs: online evaluators run "automatically on production traces: safety checks, format validation, quality heuristics, and reference-free LLM-as-judge".
- https://docs.langchain.com/langsmith/observability recorded 2026-09-29
Observability docs: "All options include observability, evaluation, prompt engineering, and deployment." across cloud, hybrid or self-hosted.
- https://www.langchain.com/langsmith/observability recorded 2026-09-29
Product page: "LangSmith offers managed cloud, bring-your-own-cloud (BYOC), and self-hosted options for teams with data residency requirements" and "Native tracing for popular agent frameworks and OpenTelemetry" plus "Online LLM-as-judge and code evals" and cost breakdowns with webhook or PagerDuty alerts.
Verified 2026-09-29