Langfuse
Open-source platform for tracing, evaluating, and improving LLM applications and AI agents.
About Langfuse
Langfuse is an open-source AI engineering platform for teams that build LLM applications and agents. It brings traces, prompt management, evaluations, datasets, and production analytics into one workflow. Developers can capture LLM and non-LLM calls with Python or JavaScript SDKs, OpenTelemetry, an LLM gateway, or more than 100 framework integrations. Traces expose latency, cost, retrieval steps, API calls, sessions, users, and agent graphs, which helps teams investigate failures and compare changes. Prompt versioning supports collaboration and deployment, while evaluation tools measure output quality and monitor production health. Langfuse Cloud offers separated US, EU, Japan, and HIPAA regions, and teams that need infrastructure control can self-host the open-source platform. The choice mainly affects operations: Cloud reduces maintenance but uses subscription pricing, while self-hosting shifts deployment, upgrades, storage, and reliability work to your team.
Best for
Best use cases
Debug agent and LLM behavior with production traces.
Version prompts and test model or prompt changes collaboratively.
Monitor quality, latency, and cost after deployment.
Compare before choosing
Choose Cloud when reduced infrastructure work matters more than self-hosting control.
Choose self-hosting when infrastructure control justifies ongoing operations work.
Check region requirements early because accounts and data are separated by region.
Features
Trace LLM applications
Capture LLM calls, retrieval, embeddings, API calls, sessions, users, and agent graphs to investigate behavior, cost, and latency.
Manage prompts
Version, test, collaborate on, and deploy prompts from a shared workflow.
Evaluate output quality
Run evaluations, monitor production health, and test application changes before wider release.
Pros
- Open-source and self-hostable for teams that need infrastructure control.
- Broad instrumentation support reduces compatibility friction and vendor lock-in.
- Integrated tracing, prompts, and evaluation shorten the debugging workflow.
Cons
- Self-hosting requires ongoing deployment, upgrade, storage, and reliability work.
- Moving between Cloud regions requires a new account and data migration.
- Hosted deployment adds recurring subscription cost.
Pricing Plans
Langfuse Cloud
- Managed cloud deployment
- Tracing, prompt management, and evaluation
- Choice of separated supported data regions
* Pricing information is for reference only. Please refer to the latest announcement on Langfuse's official website.