Used by 21 of the Fortune 50
90B+observations/month
100,000+engineers building on Langfuse
Open Source
Agent Evals &
Observability
Trace, evaluate, and improve AI agents with one open platform. Use production data to understand behavior, collaborate on fixes, and ship better quality at lower cost and latency.
Trace every LLM call with cost and latency.Watch demo
Launch, observe, improve — repeat.
Langfuse connects tracing, monitoring, datasets, experiments, and evaluation in one continuous loop. Use production signals to understand behavior, test improvements, and ship better agents with confidence.
The full AI engineering loop
See how observability, prompts, evals, experiments, and human feedback work together.
Learn in Academy01PromptVersion & deploy
02ObserveTrace production
03EvaluateScore quality
04ImproveExperiment & ship
Production data
All the tools, one
integrated platform.
One integrated platform to trace, manage prompts, evaluate, and experiment from prototype to production scale.
Observability
Hierarchical traces capture every LLM call, tool invocation, and retrieval step. Filter by user, session, cost, latency, or custom metadata.
Evaluation
LLM-as-a-judge, heuristic functions, or human review. Run evaluators on production data or during experiments.
Prompt Management
Separate prompts from code with one-click deployments and rollbacks. Turn improving your production prompts a team sport.
Playground
Test prompts on real production inputs and compare models side-by-side.
Experiments
Define test cases and run experiments. Compare results side by side.
Human Annotation
Collaborative Human-in the-Loop workflows to review traces and create golden datasets.
Cost & Latency
Monitor cost, latency, and quality with dashboards and automated alerts.
Canva's AI team relies on Langfuse to trace and debug their generative design features in production.
Start buildingWorks with any stack.
Langfuse works with any language and framework supporting OTel instrumentation. Additionally, 100+ integrations make getting started even easier. No framework lock-in.
Languages (via OTel)
TypeScript (Native SDK)Agent frameworks
Pydantic AI
Google ADK
CrewAI
LiveKitModel providers
Anthropic
Amazon Bedrock
Azure OpenAI
Mistral AI
Google Gemini
xAI
vLLM
Groq100+ more integrations
Open platform. Open source.
We are huge fans of open standards and data portability. Langfuse won't lock in your data, ever.
Docker Compose
Kubernetes (Helm)
GCP (Terraform)MIT license
- All product features MIT licensed
- Scales to billions of monthly events
- Fork, modify, contribute
APIs & exports
- REST APIs for everything
- Query SDK
- S3 blob storage export
Active OSS community
- 35,000+ GitHub stars
- 5,000+ Discord members
- Weekly releases and community hours
Made for developers, loved by agents.
Work in the app or from your IDE. Halo, the Langfuse assistant, investigates production and takes approved actions; SKILL.md, CLI, and MCP connect coding agents to Langfuse.
In-app
Halo, the Langfuse Assistant
Automate the AI engineering loop: investigate production data, understand what happened, and turn findings into approved actions without leaving Langfuse.
Debug tracesFind failed generations and traces with high latencyOptimize spendBreak down token spend, cost, and latencyBuild evalsCreate regression datasets and score configs
View documentationCoding agents
SKILL.md
A ready-made skill for managing prompts, traces, and evals through natural language.
ConfigureTerminalLangfuse CLI
Full API access from the terminal for agent workflows, scripts, and CI/CD.
ConfigureIDE agentsPlatform MCP Server
Structured access for IDE agents to manage prompts, query traces, and use Langfuse data.
ConfigureEnterprise scale and security.
Traditional observability handles many small spans. LLM systems run differently. Every step carries rich, verbose I/O that legacy platforms can't handle at scale. Langfuse ingests and queries LLM traces reliably at enterprise scale while following strict compliance frameworks.
Built for rich traces
- ClickHouse OLAP database
- Async ingestion via Redis queue
- S3/Blob storage for large payloads
- Edge-cached prompts
Production proven
- 65M+ SDK installs/month
- 90B+ observations processed per month
- 50,000+ companies using Langfuse
- 99.9% uptime
Control without compromise
SOC 2 Type II & ISO 27001
EU & US Data Regions
HIPAA-ready region
Hardening for Gov (soon)GDPRRegional data residency, data masking, retention controls, and self-hosting support strict compliance requirements.
Explore securityCanva's AI team relies on Langfuse to trace and debug their generative design features in production.
Talk to usWhy use Langfuse?
Langfuse is the most widely adopted open-source LLM engineering platform. Developers who value open-source and control over their data build production grade agents and LLM applications with Langfuse.
Get Started Plans from $29/monthStart improving
Start improving
your agents
in under 5 minutes.
or any other agentTracing:
Prompt: Install the Langfuse Agent Skill from github.com/langfuse/skills and use it to add tracing to this application with Langfuse following best practices.Evals:
Prompt: Install the Langfuse Agent Skill from github.com/langfuse/skills and use it to set up evals for this application with Langfuse. Guide me through choosing the right evaluation approach methods.Prompt Management:
Prompt: Install the Langfuse Agent Skill from github.com/langfuse/skills and use it to migrate the prompts in this codebase to Langfuse.Need help? — Talk to Sales·Reach out to Support
Questions & Answers
What is Langfuse?+
Langfuse is an open-source AI engineering platform that helps teams build, monitor, and improve their LLM applications. It covers the full development lifecycle with tracing, prompt management, evaluations, and analytics dashboards in one place. Langfuse is used by 50,000+ companies and processes billions of observations per month. You can try it instantly with the public demo project or choose a plan.
What does Langfuse help me with?+
Langfuse helps you debug LLM applications with detailed traces that capture every step of your AI pipeline, including agent graphs. You can manage and version prompts collaboratively, run automated evaluations, track costs and latency across models and providers, and run experiments on datasets before shipping. It also supports custom dashboards for team-wide visibility.
How does Langfuse compare to LangSmith, Braintrust, Arize, Galileo, and Datadog?+
Langfuse is an open-source AI engineering platform for tracing, evaluations, prompt management, and experiments. It is MIT licensed, OpenTelemetry native, self-hostable at the same scale as Langfuse Cloud, and model- and framework-agnostic. See the comparison pages for product-by-product differences and pricing.
Can I use just tracing without the other features?+
Yes. The SDKs are modular: instrument your app with a few lines of code and use only the capabilities you need. Tracing works independently of prompt management, evaluations, or any other feature.
What deployment options do exist?+
Langfuse is available as a managed cloud service in US and EU regions, or you can self-host it with Docker Compose, Kubernetes Helm, or Terraform templates for AWS, GCP, and Azure. The self-hosted version includes all product features under the MIT license.
Is self-hosting available?+
Yes. Langfuse is MIT licensed and can run on your own infrastructure. All product features are included with no feature gates; you pay for your infrastructure. Enterprise self-hosting adds support SLAs and advanced identity features for teams that need them.
What frameworks are supported?+
Langfuse supports 100+ integrations, including LangChain, LlamaIndex, CrewAI, OpenAI Agents, Pydantic AI, Mastra, Google ADK, Vercel AI SDK, OpenAI, Anthropic, and AWS Bedrock. It is OpenTelemetry native, so it also works with existing OTel instrumentation.
What's the latency impact?+
Langfuse is async by default. Tracing never blocks your application; SDKs batch and send data in the background. For prompt management, edge caching keeps prompt fetches fast and guaranteed availability keeps your application working if Langfuse is temporarily unreachable.
Is Langfuse secure and compliant?+
Langfuse Cloud is designed for enterprise workloads with SOC 2 Type II and ISO 27001 alignment, GDPR support, encryption in transit and at rest, and regional hosting options. Self-hosting gives you full control over data residency and infrastructure.
How do I get started?+
Choose a plan to create your workspace, then follow the tracing quickstart. To work with a coding agent, install the Langfuse Agent Skill and ask it to add tracing, set up evaluations, or migrate prompts.
How does pricing work?+
Langfuse Cloud offers Core, Pro, and Enterprise plans. Paid plans include monthly billable units and use graduated pricing as volume grows. See the pricing page for the calculator, limits, and feature comparison.