LangSmith and Langfuse differ on lock-in posture: proprietary + LangChain-coupled versus open-source + OTel-based; that axis dwarfs storage or sync-async differences.
Imagine two doorbell apps. One only works with one brand of doorbell, runs on the company's servers, and you cannot inspect the code; if the company changes its pricing, you change vendors or you pay. The other works with any doorbell, can run on your own machine, and the code is open; if the company changes, you keep running it yourself. The two apps may show you the same notifications today, but they leave you in very different places tomorrow. LangSmith is the first; Langfuse is the second.
Concept explanation~2 min read
Everything you need to truly understand this topic: intuition, mechanics, step by step explanation, code, formulas, and worked example. Click to expand.
Concept explanation~2 min read
Everything you need to truly understand this topic: intuition, mechanics, step by step explanation, code, formulas, and worked example. Click to expand.
Observability vendor selection looks like a tool comparison and turns out to be an ecosystem commitment. LangSmith and Langfuse cover the same surface area (spans, prompts, completions, costs, evals, datasets, prompt management) and look indistinguishable on a feature checklist. The architecturally important difference is invisible on the checklist: it lives in the licensing, the hosting model, and the ingest-path architecture.
This deep dive walks the structural commitments each vendor makes, the OpenTelemetry GenAI angle that defines the modern split, why the surface-level distinguishers (storage engine, sync vs async, prompt hashing) are red herrings, where Arize Phoenix and Helicone fit alongside the two named vendors, and the ejection mechanics that ultimately decide whether the lock-in posture matters to your team.
LangSmith: closed, hosted, LangChain-coupled
LangSmith is the tracing and evaluation product built and operated by LangChain Inc. The core service is closed-source and hosted on LangChain's infrastructure (with a self-hosted tier available for enterprise customers with compliance constraints). The ingest path is, by design, the LangChain callback handler: register the LangSmith handler in your config['callbacks'] and every Runnable, LLM call, tool call, and agent step shows up as a span in the LangSmith UI.
Non-LangChain code can be instrumented via LangSmith's @traceable decorator (Python) or traceable() wrapper (TypeScript), which is a manual instrumentation API similar in spirit to OTel's @instrument. This works but is second-class in two senses: (1) the integration is less seamless than the framework-level callback hook, and (2) the eval and dataset features assume LangChain-shaped trace structures, so non-LangChain spans show up as flat lists of LLM calls rather than nested chains.
The commercial story is also a commitment. LangSmith pricing scales with trace volume, and the only export path is via the LangSmith API. A team that has built six months of trace history, evals, and prompt experiments in LangSmith and then decides to migrate faces real work: extract historical data through the API, re-instrument code with a different vendor's SDK, rebuild dashboards. The work is bounded but not trivial.
None of this makes LangSmith a bad pick. It is the sharpest tool for teams that are committed to LangChain and want the tightest possible integration. It is a bad pick for teams that value portability, because every commitment compounds the eventual ejection cost.
Situations where this technique stops working.
2–4 min · Everything important, quickly.
Real products, models, and research that use this idea.
- Teams that committed to LangChain plus LangSmith early (2023-2024) report that ejection in 2026 means rewriting both the framework AND the observability stack at once, the cost is real.
- Klarna's customer-support stack picked Langfuse explicitly to keep the OTel exit open as they migrated from LangChain to thin SDK wrappers on critical paths.
What an interviewer would ask next. Try answering before peeking at the approach.
QWhat is the eject from LangSmith story when a team decides to migrate?
Two parts: (1) export historical traces from LangSmith (their API supports this but it is a project, not one-shot); (2) re-instrument code with the new vendor's SDK or OTel exporter, which means touching every chain config call site. The historical-data part is what makes the migration heavy; the re-instrumentation is bounded.
Red flags & common mistakes
The phrases that signal junior thinking. Click to expand.
Red flags & common mistakes
The phrases that signal junior thinking. Click to expand.
Picking the vendor on dashboard polish or eval-feature richness without checking the licensing and OTel posture. Today's feature gap closes in months; lock-in posture is permanent.
60 second bullets to scan on the way to the call.
LangSmith's licensing and hosting model
Langfuse's licensing and hosting model
Primary sources. Browse if you want the original framing.
Same topic, related formats. Practice these next.