Langfuse
Open-source LLM observability and analytics
Overview
Langfuse is the leading open-source LLM observability platform — traces, evaluations, prompt management, costs and analytics for LLM applications. Self-host (MIT licensed) or use the cloud-hosted SaaS. Strong adoption among teams running LLM products at production scale.
Pricing
Pricing shown for reference only. These figures reflect RECATOOLS research as of 19 May 2026 and may be out of date or incomplete. This is not financial or purchasing advice — always confirm the current price on the provider’s official website before making any decision.
Use cases
What you can produce with Langfuse
- Trace every step of an LLM application — nested agent calls, tool invocations, retrievals — with latency, token counts and cost per step.
- Self-host the full MIT-licensed platform on your own infrastructure so LLM telemetry never leaves your environment.
- Manage, version and deploy prompts from the Langfuse UI so prompt changes ship without a code redeploy.
- Run LLM-as-judge evaluators over production traces to score outputs for quality, hallucination or policy violations automatically.
- Build regression-test datasets from real production traces and run experiments comparing models or prompt versions against them.
- Break down LLM spend by user, feature, model or API key on cost dashboards to find what is burning budget.
- Attach end-user feedback (thumbs up/down, scores, comments) to specific traces so bad outputs can be traced back to their exact inputs.
ASEAN Perspective
Langfuse in Southeast Asia
ASEAN-region availability and pricing notes coming soon. Drop the editorial team a note via /contact/ if you can supply local context (Singapore/Malaysia/Indonesia/Thailand/Vietnam).
Langfuse is a leading open-source LLM observability platform covering tracing, prompt management, evaluations, cost tracking and analytics, with framework-agnostic SDKs that work well alongside LangChain, LlamaIndex or raw API calls. It suits engineering teams who want production-grade LLM monitoring with the option to self-host for data control rather than being locked into a vendor cloud.
It is one of the strongest tools in its niche; the main caveats are that you take on ops overhead if self-hosting, and the breadth of features means some setup investment. Globally available, transparent open-core pricing, well-documented SDKs. No SEA-specific residency, but self-hosting lets regional teams keep data local.
What people say
Langfuse's status changed materially in January 2026: it was acquired by ClickHouse, announced alongside ClickHouse's $400M Series D at a $15B valuation. The companies have committed to keeping the core MIT-licensed and self-hostable, with Langfuse Cloud continuing as a standalone service — and as of mid-2026 that promise is holding, though open-source users are watching for feature gating. By the acquisition, Langfuse claimed 19 of the Fortune 50 among users, 2,000+ paying customers and tens of millions of SDK installs per month, making it the most widely deployed open-source LLM observability platform.
User sentiment across Hacker News, Reddit and dev communities is strongly positive. The recurring praise: you can self-host production-grade LLM tracing without a per-seat enterprise contract, the nested trace view makes multi-step agent debugging genuinely tractable, and the platform is framework-agnostic — it works the same whether you use LangChain, LlamaIndex, the OpenAI SDK or raw HTTP. Cost is another draw: at high trace volumes, published comparisons put Langfuse Cloud at roughly a third of LangSmith's price, and self-hosting far below both.
The criticisms are practical rather than damning. Self-hosting the v3 architecture is heavier than the old single-container setup — it wants ClickHouse, Redis and object storage, which is real ops work for small teams. The UI is dense and takes orientation time, some advanced features (SSO enforcement, certain enterprise controls) sit behind paid tiers, and teams deeply invested in LangGraph note that LangSmith's native integration there is smoother.
Langfuse fits engineering teams running LLM features in production who want ownership of their telemetry — especially those with data-residency requirements or existing infrastructure to host it. Teams wanting zero-ops, all-in-one LangChain tooling may still prefer LangSmith.
Summary of public user & expert reviews, compiled by RECATOOLS.
About this listing
This entry was compiled from publicly available data including Langfuse's official website, press releases, documentation, and reputable third-party publications. RECATOOLS is not affiliated with Langfuse unless explicitly stated.
Third-party AI tools update their pricing, features, availability, and policies frequently. Information here may be outdated by the time you read this — we make reasonable efforts to keep listings current, but cannot guarantee absolute accuracy.
For the latest details, please refer to Langfuse directly →
Spotted something out of date? Suggest an update →
Langfuse in the news
More in Code & Dev Tools
