starter.wiki

Managed LLM observability and evaluation

LangSmith

A LangChain-managed platform for tracing, evaluation, datasets, experiments, and production monitoring with Cloud and Enterprise self-hosted routes.

Visit official site

Editorial verdict

Choose LangSmith when its tracing and evaluation workflow aligns with the development stack and the team accepts its usage-unit, retention, data, and deployment boundaries.1234

Best for

  • LangChain and LangGraph application teams
  • Managed evaluation and experiment workflows
  • Organizations needing documented trace-retention controls
1234

Not ideal for

  • Teams requiring an OSS-first observability core
  • Buyers unable to model evolving usage units
  • Self-hosted teams unprepared to operate required data services
1234
Main trade-off

Framework-aligned managed evaluation reduces integration work while adding usage-unit, trace-tier, retention, platform, and Enterprise self-hosting dependence.1234

Product boundary

Whether LangSmith's tracing and evaluation workflow, especially for LangChain or LangGraph systems, fits usage, retention, deployment, and data-governance requirements.

This page evaluates LangSmith observability and evaluation. LangGraph runtimes and LangSmith Deployment are separate product decisions. Base and extended traces, datasets, Cloud plans, and Enterprise self-hosting have different retention and infrastructure boundaries.1234

For: Teams wanting managed tracing and evaluation with strong LangChain or LangGraph workflow alignment

  • The production workload, language, model, provider, or deployment boundary changes
  • Current pricing, retention, data use, policy, or regional support changes
  • Measured quality, latency, reliability, or operating cost no longer fits

Why teams consider LangSmith

  • Framework alignmentTracing and evaluation integrate naturally with LangChain and LangGraph workflows.1234
  • Evaluation workflowDatasets, experiments, evaluators, and production traces connect development and monitoring.1234
  • Retention controlsOfficial administration docs distinguish base traces, extended traces, and datasets.1234

Pricing

Seat and usage-based Cloud plans use current LangSmith usage units and trace classes; Enterprise and self-hosted terms are custom.1

Current official route

Developer, Plus, or Enterprise

Verified 2026-07-27: LangSmith publishes Developer and Plus seat or usage routes and custom Enterprise terms. Current usage-unit and trace-retention definitions must be used instead of legacy trace pricing.1

Cloud cost
Seats, included usage, current usage units, and trace class1
Retention
Base and extended traces have different retention; datasets persist separately1
Self-hosted
Enterprise add-on plus customer-operated data services1
Pricing checked View official pricing

LangSmith vs alternatives

Langfuse

Choose when
Choose Langfuse when open-source portability or self-hosting materially affects the observability decision, after modeling ingestion units, retention, storage, evaluation models, and platform operations.
Avoid when
Teams unwilling to operate self-hosted data infrastructure
Compared with LangSmith
Open-source deployment choice increases control and portability while adding ingestion, retention, storage, ClickHouse, upgrade, evaluation-model, and edition responsibility.56

Braintrust

Choose when
Choose Braintrust when evaluation-led development is the operating model and its processed-data, scoring, retention, collaboration, and deployment boundaries fit.
Avoid when
Teams primarily seeking an OSS tracing backend
Compared with LangSmith
A focused evaluation workflow improves repeatability while adding dataset, scoring, processed-data, retention, collaboration, and hosted-platform dependence.78

Arize Phoenix

Choose when
Choose Arize Phoenix when an open-source and OpenTelemetry-oriented tracing and evaluation stack justifies owning deployment, storage, access, upgrades, and evaluation-model cost.
Avoid when
Teams wanting a fully managed default
Compared with LangSmith
Open-source tracing and evaluation maximize inspectability and control while transferring data infrastructure, access, retention, upgrade, evaluation-model, and support responsibility.910

Resources and sources

Official product, documentation, pricing, and policy sources

  • LangSmith pricing
    Open
  • LangSmith administration and retention
    Open
  • LangSmith data purging
    Open
  • Self-hosted LangSmith
    Open
  • Langfuse Cloud pricing
    Open
  • Langfuse self-hosted pricing and editions
    Open
  • Braintrust pricing
    Open
  • Braintrust tracing and instrumentation
    Open
  • Arize Phoenix documentation
    Open
  • Arize Phoenix self-hosting configuration
    Open

Related tasks

  1. 1
    LangSmith pricing

    LangChain · Accessed Official

  2. 2
    LangSmith administration and retention

    LangChain · Accessed Official

  3. 3
    LangSmith data purging

    LangChain · Accessed Official

  4. 4
    Self-hosted LangSmith

    LangChain · Accessed Official

  5. 5
    Langfuse Cloud pricing

    Langfuse · Accessed Official

  6. 6
    Langfuse self-hosted pricing and editions

    Langfuse · Accessed Official

  7. 7
    Braintrust pricing

    Braintrust · Accessed Official

  8. 8
    Braintrust tracing and instrumentation

    Braintrust · Accessed Official

  9. 9
    Arize Phoenix documentation

    Arize · Accessed Official

  10. 10
    Arize Phoenix self-hosting configuration

    Arize · Accessed Official