Last updated 08-09-2026
Category:
Reviews:
Join thousands of AI enthusiasts in the World of AI!
Langfuse
Langfuse traces LLM applications so you can see every prompt, model call, tool step, and cost in one place. It captures agent runs, chat sessions, and multi-modal inputs through Python and JavaScript SDKs or OpenTelemetry, then links them to prompt versions and evaluation scores. Teams use it to debug production failures instead of guessing from raw API logs.
Generic logging tools were not built for non-deterministic model chains. Langfuse treats traces, observations, and scores as first-class objects, with prompt management, dataset experiments, and LLM-as-judge evaluators in the same project. You can self-host the open-source edition or use Langfuse Cloud when you want managed hosting in US, EU, or Japan regions.
ML engineers wire Langfuse into LangChain, LiteLLM, or custom apps to track token spend. Product teams compare prompt versions before rollout, and platform groups run annotation queues for human review. The free Hobby cloud plan includes 50,000 billable units per month with no credit card required.
Trace full LLM agent runs with spans, generations, token counts, and cost breakdowns
Version prompts and deploy labels to production without code changes
Run dataset experiments and LLM-as-judge evaluators inside the same project
Python and JavaScript SDKs plus OpenTelemetry for Java, Go, and custom stacks
Hobby cloud plan includes 50,000 billable units per month with no credit card
Self-host the open-source edition with Docker Compose or Kubernetes templates
Open-source core with a generous free cloud tier for early projects.
Tracing, prompt versioning, and evaluations live in one platform.
Graduated usage pricing lowers per-unit cost as ingestion volume grows.
Advanced compliance features like SCIM and audit logs require Enterprise pricing.
Additional usage beyond included units adds cost quickly on high-traffic apps.
What does Langfuse do?
Langfuse is an LLM engineering platform for tracing, prompt management, and evaluation. It records model calls, tool steps, and scores so teams can debug and improve production AI applications.
Is Langfuse free to use?
Langfuse offers a free Hobby cloud plan with 50,000 billable units per month and an open-source self-hosted edition. Paid Core plans start at $29 per month for longer data retention.
Can I self-host Langfuse?
Yes. Langfuse is open source and can be self-hosted with Docker Compose or Kubernetes. The self-hosted edition includes tracing, prompts, and evaluations without a Langfuse Cloud subscription.
What is a Langfuse billable unit?
A Langfuse billable unit is any tracing data point sent to the platform, including traces, observations such as spans and generations, and evaluation scores.
Which SDKs does Langfuse support?
Langfuse provides Python and JavaScript SDKs plus OpenTelemetry integrations for Java, Go, and custom stacks. It also integrates with LangChain, LiteLLM, and other frameworks.
How much does Langfuse Pro cost?
Langfuse Pro is $199 per month on cloud and includes 100,000 billable units, three years of data access, SOC2 reports, and high API rate limits.