← All companies

Agents / Company profile

Langfuse

What Langfuse does

Langfuse is an open-source AI engineering platform for building production-grade LLM applications through end-to-end observability and quality management. It combines (1) tracing and analytics for LLM app execution, (2) prompt management with monitoring, (3) evaluations for measuring output quality (including “LLM-as-a-judge” style evaluations), and (4) datasets and workflows that support testing and continuous improvement.

Langfuse serves two main audiences: development teams integrating LLM applications (who use the open-source SDKs/instrumentation and the UI to debug production behavior) and product/engineering teams running evaluation pipelines over historical and live traces to monitor quality, latency, and cost. The company offers both self-hosted (open-source) deployments and a hosted “Langfuse Cloud” offering with usage-based units and enterprise features such as audit logs, SCIM, uptime/SLA support, and security controls.

In January 2026, Langfuse was acquired by ClickHouse, and its roadmap and platform evolution increasingly align with ClickHouse’s analytics stack. A recent example is Langfuse v4, which (per Langfuse’s changelog) is designed for faster and more scalable querying and dashboards by changing the underlying data model and query behavior. Langfuse’s current strategic position is therefore best described as an open-core platform (open-source instrumentation/data model plus commercial cloud/enterprise features) that is now integrated into ClickHouse’s broader OLAP and AI analytics ecosystem.

News

Company record

Aug 21, 2026 · Official · Langfuse ChangelogLangfuse CLI 1.0

Langfuse announced Langfuse CLI 1.0, describing a rewritten engine with faster invocations and CLI changes such as plain flags for union endpoints and exit codes intended to be machine-readable by agents.

Aug 19, 2026 · Official · Langfuse ChangelogTurn scores into charts

Langfuse added the ability to convert filtered score-table views into charts and to spot outliers via a “Pulse” strip that links back to underlying rows.

Aug 18, 2026 · Official · Langfuse ChangelogCost tracking support for OpenAI Fast mode

Langfuse added automatic cost tracking for OpenAI Fast mode, including Fast/priority tier matching and long-context rate handling for supported GPT-5.6 calls.

Aug 17, 2026 · Official · Langfuse ChangelogLangfuse v4 is live: faster at scale, with more ways to search, monitor, and evaluate

Langfuse announced that Langfuse v4 is live on Langfuse Cloud and generally available for self-hosted deployments, highlighting faster initial loads for large datasets and improved ways to search, monitor, and evaluate agent steps.

Aug 11, 2026 · Official · Langfuse ChangelogInstance switcher for self-hosted deployments

Langfuse introduced an enterprise self-hosted UI feature to switch between multiple Langfuse deployments from the sidebar user menu.

May 31, 2026 · Official · Langfuse BlogLangfuse May Update (Code Evaluators, full-text search, Langfuse MCP, Experiments in CI/CD)

Langfuse’s May update announced multiple feature drops including code evaluators (deterministic Python/TypeScript checks), full-text search, an expanded Langfuse MCP surface, and running dataset-based experiments in GitHub Actions CI/CD.

Jan 16, 2026 · Official · Langfuse BlogLangfuse joins ClickHouse (acquisition announcement)

Langfuse announced it was acquired by ClickHouse and discussed implications for users and the shared history between the companies (including Langfuse running on ClickHouse).

Jan 16, 2026 · Official · ClickHouse BlogClickHouse welcomes Langfuse: The future of open-source LLM observability

ClickHouse announced the acquisition of Langfuse and described it as an open-source platform for LLM observability, evaluations, and prompt management.

Show 1 earlier updates
Source map · 5 recurring channels · 10 references

Still resolving: Newsroom · Official X

What it builds

Langfuse Cloud

Hosted Langfuse for LLM observability and evaluation, with pricing plans expressed in included monthly units and an enterprise tier with additional security and support features.

source ↗
Self-hosted Langfuse (open-source)

A self-hosted deployment option for Langfuse (with an open-source core) so teams can run Langfuse in their own infrastructure while retaining access to the same observability and evaluation capabilities.

source ↗
Langfuse Observability (Tracing)

Tracing and execution analytics for LLM applications, designed to capture execution context and enable debugging and production monitoring across model calls, tool calls, and other app interactions.

source ↗
Prompt Management

Prompt version control and deployment workflows integrated with monitoring and evaluation so changes can be tracked back to observed behavior and quality signals.

source ↗
Evaluations (LLM-as-a-Judge and datasets)

Evaluation capabilities, including LLM-as-a-judge evaluations and the ability to build structured evaluation workflows over datasets and production traces.

source ↗
Langfuse MCP server

A Model Context Protocol (MCP) server intended to expose Langfuse capabilities (including prompts and other operational data/actions) to MCP clients and agents.

source ↗

Milestones & partnerships