SoftBestsLangfuse
Open-source tracing, evaluation and prompt management for LLM apps in production.
Best for
Teams shipping LLM features who need to see what every model call did, cost and why it failed.
Why it deserves attention
Langfuse records each LLM request as a trace with nested spans, prompts, outputs, token counts and latency, so you can debug a bad answer or a runaway cost by opening one trace instead of grepping logs. It adds prompt versioning and a playground, datasets and automated evaluations, and dashboards for cost and quality over time. It works with OpenTelemetry and has SDKs for Python, TypeScript and popular frameworks. Self-hosting is one Docker Compose stack with Postgres, ClickHouse and Redis; the cloud plan has a usable free tier. It has roughly 37,000 GitHub stars and an active weekly release cadence. The caveat: self-hosting is a real multi-service deployment, and evaluation features require some setup discipline.
Ratings & reviews
0 ratings0 people rated this
No ratings yet. Be the first to rate this tool.
Can replace
- LangSmith
Related and alternatives
More in Agent infrastructure
Underratedagent-browser
A native Rust CLI that gives coding agents a browser they can drive one short command at a time.
UnderratedBrowserbase
Managed headless browsers with sessions, proxies, and observability for web agents.
UnderratedCindy
An open-source AI agent that connects your computers and phone into one workspace.