Langfuse: Debug and improve LLM apps with production data
Langfuse is an open-source LLM engineering platform for tracing, prompt management, evaluation and analytics of LLM applications and agents.
What is Langfuse?
Langfuse records hierarchical traces of LLM calls, tool invocations and retrieval steps, and lets teams filter them by user, session, cost, latency or custom metadata. Around that tracing data it adds prompt versioning and release management, a playground for testing prompts on real inputs, datasets and experiments, LLM-as-a-judge and human annotation workflows, plus dashboards and alerts for cost and latency. The platform is open source under the MIT license and can be self-hosted with Docker Compose, Kubernetes via Helm, or Terraform on AWS, GCP and Azure. It is OpenTelemetry native, ships Python and TypeScript SDKs, and lists more than 100 integrations covering agent frameworks, model providers and tools. A hosted cloud option is also available with US, EU or JP data regions.
Focus areas: Prompt Management.
Who is Langfuse for?
- Engineering teams shipping LLM applications and agents
- Platform and ML engineers who need tracing and evaluation in one place
- Teams that require self-hosting or data residency control
How much does Langfuse cost?
Langfuse starts at $29/mo. Always check the vendor’s own pricing page before you buy.
| Plan | Price | What you get |
|---|---|---|
| Hobby | Free | 50k units / month, 30 days data access, 2 users |
| Core | $29 / month | 100k units / month, 90 days data access, unlimited users |
| Pro | $199 / month | 100k units / month, 3 years data access, high rate limits |
| Teams Add-on | $300/mo | Enterprise SSO, SSO enforcement, fine-grained RBAC |
| Enterprise | $2499 / month | Everything in Pro + Teams, audit logs, SCIM API, custom rate limits |
Is Langfuse worth it?
A broad, open-source option for teams that want tracing, prompts and evaluation in one place without framework lock-in
Where Langfuse is strong
- Combines tracing, prompt management, evaluation and analytics in one platform
- Open source under MIT with documented self-hosting options
- OpenTelemetry native with Python and TypeScript SDKs and 100+ integrations
Where it falls short
- Cloud plan pricing is usage-based beyond the included units, so costs scale with observation volume
- The source does not state a free self-hosted support entitlement or a published self-hosted support policy
How do you get started with Langfuse?
- 01Sign up for the free Hobby tier, which includes 50k units per month and needs no credit card
- 02Install the Python or TypeScript SDK, or send OpenTelemetry data from another language, and start tracing calls
- 03Review traces in the dashboard, then add prompt management, datasets and evaluators as needed
Langfuse: frequently asked questions
What does Langfuse actually do?
It traces LLM calls, tool invocations and retrieval steps, and adds prompt management, evaluation, experiments, human annotation and cost and latency dashboards on top of that data.
Is there a free plan?
Yes. The Hobby plan is free with no credit card required and includes 50k units per month, 30 days of data access and 2 users.
Can I self-host Langfuse?
Yes. The product is MIT licensed and can be self-hosted using Docker Compose, Kubernetes with Helm, or Terraform on AWS, GCP and Azure.
Which frameworks and languages are supported?
Langfuse works with any language or framework that supports OpenTelemetry instrumentation, has native Python and TypeScript SDKs, and lists over 100 integrations including LangChain, Vercel AI SDK, LiteLLM and OpenAI.
Langfuse alternatives
The closest tools in our directory, matched on category, focus areas and audience.
| Tool | Best for | Free plan | Starting price | Focus areas |
|---|---|---|---|---|
| Langfuse | Teams tracing, evaluating and improving LLM apps in production | Yes | $29/mo | Prompt Management |
| Opik | Developers tracing, evaluating and monitoring LLM agents | Yes | — | Prompt Management |
| Lunary | Open-source platform to monitor, manage and improve your LLM apps. | Unknown | — | Prompt Management |
| Promptitude.io | Teams managing and deploying prompts across multiple AI models | Yes | $39/mo | Prompt Management |
| Langtail | Teams that need shared, testable prompt management for LLM apps | Yes | $99/mo | Prompt Management |
| PRST.ai | Teams managing prompts and AI connectors on their own infrastructure | Yes | — | Prompt Management |
| Cleanlab | An open-source framework for data-centric AI that automatically finds and fixes errors in datasets. | Yes | — | — |
| LangWatch | LLM observability and evaluation platform for monitoring, evaluating, and optimizing LLM applications. | Yes | — | — |
| Giskard | An open-source framework for testing all ML models, from tabular to LLMs. | Yes | Free | — |
| UpTrain | Teams evaluating, testing and monitoring LLM apps in production | Unknown | — | — |
| WhyLabs | Teams studying open source AI observability and data logging | Unknown | — | — |

