Source: LangSmith tracing SDK, OTel export and deployment pricing

LangSmith docs on @traceable, wrap_openai, trace/RunTree and flushing, OTel environment variables, plus the pricing page's plans and Agent Server metering and the 2025 product rename.

Contents

Summary

How code that is not LangChain gets traced into LangSmith, how traces leave the process, and what the hosted platform costs and is called after the October 2025 rebrand.

Key claims

  • @traceable turns any function into a run (nested calls become children); run_type="llm" enables token/latency display. wrap_openai traces an OpenAI client. Python’s trace context manager and the low-level RunTree give manual control; RunTree posts regardless of LANGSMITH_TRACING.
  • tracing_context(enabled=...) toggles tracing at runtime. LangChain and LangGraph are traced automatically when LANGSMITH_TRACING=true and LANGSMITH_API_KEY are set.
  • Traces are sent from a background thread; a short-lived process can exit before upload, so call client.flush() in a finally block.
  • OTel: LANGSMITH_OTEL_ENABLED exports LangChain/LangGraph spans via OTel; LANGSMITH_OTEL_ONLY sends only to your OTel endpoint. Ingestion is asynchronous and returns 200 before processing.
  • Pricing (page read 2026-11-14): Developer free (1 seat, up to 5k base traces/month), Plus $39/seat/month (unlimited seats, 10k base traces/month); both Developer and Plus include one small serverless Agent Server at no charge, Enterprise custom with hybrid or fully self-hosted options. Base traces retained 14 days, extended 180 days (upgrade fee). Additional Agent Servers metered in LSUs (1 LSU = $1): 0.0675 LSU per runtime vCPU-hour, 0.009 per runtime GiB-hour, 0.177 per database vCPU-hour, 0.025 per database GiB-hour.
  • October 2025: LangGraph Platform renamed LangSmith Deployment and LangGraph Studio renamed LangSmith Studio; naming change only, APIs and pricing unchanged.

Feeds

LangSmith, LangGraph

Suggest changes →