
LLM engineering platform — tracing, prompt management, cost tracking.
Langfuse is an open-source LLM engineering platform with 28k+ GitHub stars — a LangSmith alternative for tracing, debugging, and monitoring AI applications.
Compare Langfuse with Braintrust and Arize Phoenix before you choose your stack.
| Langfuse | LangSmith | |
|---|---|---|
| License | MIT (open source) | Proprietary |
| Data residency | Self-hosted option | Vendor cloud |
| Privacy | No third-party tracking by default | Vendor analytics pipeline |
| Customization | SQL and plugin extensibility | Limited to product UI |
| Cost | Free tier + self-hosting | Usage-based SaaS |
Choose Langfuse if you want open-source code, self-hosting options, and full control over your data and deployment.
Choose LangSmith if you prefer a managed proprietary product with vendor support and minimal setup.
Browse more open-source alternatives to LangSmith, or explore other tools in Analytics.
| License | MIT |
| Stack | TypeScript, Python |
| Self-hosted | Yes |
| Cloud | cloud.langfuse.com |
| SDKs | Python, TypeScript, Go |
docker compose up -d
Requires PostgreSQL and ClickHouse (or Postgres for small deployments).
Yes. Langfuse is open source under MIT. You can self-host it at no software cost — you only pay for infrastructure or optional managed services.
Langfuse gives you source code access, self-hosting, and data ownership. LangSmith is a proprietary product focused on managed convenience. See the comparison table above for a side-by-side breakdown.
Yes. Langfuse supports self-hosted deployment, which is a core reason teams choose it over LangSmith. Check the Getting started or Self-hosting section for install commands.
Langfuse is actively maintained with a strong open-source community. Many teams run it in production as a Analytics alternative to LangSmith. Review the At a glance table for license and stack details.
Browse alternatives to LangSmith for more open-source options, including tools compared to Braintrust. Explore the full Analytics category for related projects.

Langflow
Visual builder for AI agents and RAG pipelines — LangChain-powered.
Helicone
LLM observability and AI gateway — one-line integration for logging, cost tracking, prompt management, and rate limiting

Laminar
Agent observability and evaluation platform — signal alerts, trace debugging, and automated evals for LLM products
mlop
MLOps platform with W&B-compatible API — experiment tracking, model registry, and collaboration for ML teams