Langfuse
Langfuse is an open-source LLM engineering platform that gives teams tracing, evaluation, prompt management, and dataset tooling to debug and improve LLM-powered applications. It can be run as a managed cloud service or self-hosted on a team's own infrastructure.
What gets deployed with it
- PostgreSQL HA Cluster — required Primary store for Langfuse trace, span, score, and evaluation metadata.
- ClickHouse Cluster — required Analytics store for trace and event queries. Sized for compressed trace data derived from daily_traces × trace_retention_days.
- Redis Cluster — required Queue and cache layer for Langfuse background worker ingestion and operational support.
Baseline topologies
| Topology | Minimum nodes | Roles |
|---|---|---|
| single-node | 1 | langfuse |
| app-plus-workers | 2 | langfuse-web, langfuse-worker |
What drives the cost
- Daily traces or events
- Approximate LLM traces or events ingested per day across all projects.
- Trace retention period (days)
- How long traces and spans stay available before expiry. Drives total storage footprint.
Where it runs
bare_metal, dedicated, cloud_vm
Size Langfuse and get a cost estimate