LangSmith

## Editorial notes

LangSmith is useful for teams that need traceability, regression testing, and evaluation around LLM and agent systems.

## Key capabilities

- **Tracing** — captures detailed traces of LLM and agent runs (inputs, outputs, tool calls, latency, token usage).
- **Evaluation** — run dataset-based and LLM-as-a-judge evaluations, plus regression testing across prompt/model versions.
- **Prompt management** — version, compare, and test prompts in a playground.
- **Framework integration** — native LangChain/LangGraph support, plus an SDK and OpenTelemetry for other stacks.

## How LangSmith compares

LangSmith competes with several LLM observability/evaluation tools in this directory:

| Tool | Type | Self-hostable | Notable for |
| --- | --- | --- | --- |
| **LangSmith** | Observability + evaluation platform | Enterprise tier | Deep LangChain / LangGraph integration |
| **Arize Phoenix** | Open-source observability + evals | Yes | OpenTelemetry-based tracing that runs locally |
| **Helicone** | Observability via a request-logging proxy | Yes | One-line integration that captures requests |
| **Braintrust** | Evals + experimentation | SaaS | Eval-first experimentation workflow |

Pick LangSmith if you are building on LangChain/LangGraph and want tightly integrated tracing and evals; consider Phoenix for an open-source, self-hostable alternative or Helicone for proxy-based request logging.

## Disclosure

Editorial listing. No paid placement or affiliate link is used.