About Langfuse
Langfuse is an open platform for tracing, evaluating and monitoring AI applications. It can be used as a cloud service or hosted on your own servers.
What it does
Langfuse records the calls an AI application makes, including the steps inside an agent. These records show usage, cost and timing, and they can be scored with evaluations. Teams use the same production data to find problems, discuss fixes and test changes before release. Instrumentation works through its SDKs or through OpenTelemetry, and some model calls can be routed through its AI gateway without changing application code.
Key features
- Trace model calls and agent steps
- Track usage and costs
- Run evaluations on production data
- Manage prompts in one place
- Route model calls through an AI gateway
- Self-host with Docker Compose, Kubernetes or Terraform
Who it's for
AI engineering teams, platform teams and product teams that run LLM features in production. Companies with strict data rules may prefer the self-hosted option.
Pricing and good to know
The project is open source and can be self-hosted. A managed cloud version is also available, and the site offers a free way to start. Check the pricing page for current plan limits. Self-hosting means your team runs and updates the servers, so plan for that work.
Highlights
- Trace model calls and agent steps
- Track usage and costs per request
- Run evaluations on production data
- Manage prompts in one place
- Route model calls through an AI gateway
- Self-host with Docker Compose, Kubernetes or Terraform
