Langfuse
FreeAn open-source LLM engineering platform for tracing, evaluation, prompt management, and metrics. [#opensource](https://github.com/langfuse/langfuse)
About Langfuse
Langfuse is an open-source LLM engineering platform that provides observability, evaluation, prompt management, and experimentation for AI agents. It enables teams to trace every LLM call, tool invocation, and retrieval step; evaluate with LLM-as-a-judge, heuristics, or human review; manage prompts with versioning and rollbacks; and run experiments on production data. Used by 19 of Fortune 50 companies and over 100,000 engineers, Langfuse supports self-hosting or cloud deployment, integrates with 100+ tools (LangChain, OpenAI, Anthropic, etc.), and scales to billions of observations per month.
Key Features
Hierarchical observability with traces for every LLM call, tool invocation, and retrieval step
Evaluation with LLM-as-a-judge, heuristic functions, and human review
Prompt management with versioning, deployment, and rollbacks
Playground for testing prompts on production inputs and comparing models
Experiments with test cases and side-by-side result comparison
Human annotation workflows for collaborative review and golden datasets
Cost and latency monitoring with dashboards and automated alerts
100+ integrations with frameworks, providers, and tools
Self-hostable with Docker, Kubernetes, Terraform, MIT license
Pros & Cons
Pros
- Open source with MIT license and self-hosting option, no vendor lock-in
- Comprehensive platform combining tracing, evaluation, prompt management, and experiments
- Wide integration ecosystem supporting major frameworks and model providers
- Used and trusted by Fortune 50 companies and large scale deployments
- Active open source community with 31k+ GitHub stars
- Scales to billions of observations per month
Cons
- Advanced features and higher usage limits require paid plans (Core $29/mo, Pro $199/mo, Enterprise $2499/mo)
- Self-hosting requires infrastructure setup and maintenance expertise
- Some features like SSO, RBAC, audit logs only available on higher tier plans
Best For
Debugging and improving LLM applications in productionTesting and comparing prompt variations before deploymentMonitoring cost and performance of AI agents at scaleCollaborative evaluation and annotation of LLM outputsBuilding production-grade LLM applications from prototype to scale
Alternatives to Langfuse
FAQ
Is Langfuse open source?
Yes, Langfuse is open source under the MIT license. All product features are MIT licensed.
Can I self-host Langfuse?
Yes, Langfuse supports self-hosting via Docker Compose, Kubernetes (Helm), and Terraform for AWS, GCP, and Azure.
What integrations does Langfuse support?
Langfuse supports 100+ integrations including LangChain, Vercel AI SDK, LiteLLM, Pydantic AI, OpenAI, Anthropic, Bedrock, and many more via OpenTelemetry.
Is there a free plan?
Yes, the Hobby plan is free with up to 50k units per month, 30 days data access, 2 users, and community support.
How does pricing work?
Pricing is based on units per month. Hobby is free, Core is $29/mo for 100k units, Pro is $199/mo, Enterprise is $2499/mo. Additional units cost $8/100k units.