Langfuse
Open source LLM observability, prompt management and evaluation platform
About Langfuse
Langfuse is an open source platform for tracing LLM and agent applications, managing prompts and running evaluations. It is available as managed cloud or self-hosted and is now part of ClickHouse.
Common use cases
- Debugging LLM apps
- Prompt iteration
- Quality evaluation
- Cost monitoring
Social listening
What people are saying about Langfuse
Reviewers like the trace debugging, prompt management and self-hosting option, and buyers typically plan for query performance at high trace volumes and evaluation setup effort; Langfuse was acquired by ClickHouse, which drew substantial Hacker News attention.
What stands out
- Reviewers find trace visualization useful for debugging multi-step LLM pipelines such as retrieval, reranking and generation.
- Central prompt management is seen as tidier than keeping prompts in config files or notebooks.
- Setup is described as quick and it fits into existing AI stacks.
- The open-source, self-hostable model gives teams more control over sensitive data.
Things to consider
- Worth confirming query performance on large trace volumes, especially date-range filtering, as traffic grows.
- Teams with complex agents or chains typically plan time to organize the trace view.
- Plan for effort to build evaluation datasets and scoring logic.
Often compared with: Arize Phoenix, OpenObserve
Sources (2)
Themes summarized from public reviews, community discussion and press coverage as of November 2026. They reflect public discussion, not a rating or an endorsement by AI Connect.
Products and services
AI Observability
Tracing and monitoring for LLM applications.
Prompt and skill management
Versioning and management of prompts.
Evaluation
Evaluation of LLM application quality.
Metrics
Usage and quality metrics.
AI Gateway
Gateway for model access.
Pricing and editions. Open source with self-hosting supported, alongside the managed Langfuse Cloud.
Recent product updates
- Acquired by ClickHouse; the announcement says Langfuse stays open source, Cloud and self-hosting continue, and the roadmap is unchanged (early 2026, exact date not confirmed).
- Langfuse v4 changelog (August 2026) describes agent evals and traces rebuilt on a single ClickHouse table.
Key features
- LLM tracing
- Prompt versioning
- Evaluations
- Cost tracking
- Self-hosting
- Datasets and experiments
Integrations
- OpenAI
- Anthropic
- LangChain
- LlamaIndex
- AWS
- Google Cloud
- Microsoft Azure
Langfuse alternatives
FAQs
Who owns Langfuse?
It is part of ClickHouse.
Can I self-host?
Yes, via Docker Compose, Kubernetes or Terraform; core features are free under MIT.
Where is the cloud hosted?
US, EU, Japan and HIPAA-ready regions.

Compare with similar providers
All ML Operations providers- Arize AIObservability and evaluation platform for AI agents, LLMs and ML models
- CometExperiment management and the open source Opik LLM observability tool
- HoneyHiveObservability and evaluation platform for AI agents in production
- PromptLayerPrompt management, evaluation and observability for LLM applications
- Actualyze AIAI governance gateway between applications and model providers
- ArthurControl plane for governing, monitoring and securing enterprise AI agents
- Blackstraw.aiAI and data services firm with accelerators for vision, GenAI and data
- BraintrustEvaluation and observability platform for AI agents
Become a Sponsor



