Deploy Langfuse
Open-source LLM tracing, evals and prompt management on Langfuse v4
Langfuse-Worker
Just deployed
Langfuse
Just deployed
Postgres
Just deployed
/var/lib/postgresql/data
ClickHouse
Just deployed
/var/lib/clickhouse
Redis
Just deployed
/data
langfuse-storage
Bucket
Just deployed
Deploy and Host Langfuse on Railway
Langfuse is an open-source LLM engineering platform. It records traces of LLM calls, agents and tool use, tracks cost and latency, manages and versions prompts, and runs evaluations, datasets and experiments. Apps send data with the Langfuse SDKs for Python and JavaScript or any OpenTelemetry exporter, and teams review it in a web UI.
About Hosting Langfuse
Langfuse v4 runs as two containers from the same release: a web service for the UI and APIs, and a worker that processes incoming events in the background. It needs Postgres for users, projects and prompts, ClickHouse for traces, observations and scores, Redis for queues and caching, and S3-compatible storage for raw events, media and exports. This template pins Langfuse 4.78.0 and connects all of them over Railway's private network, using a Railway bucket for storage. Secrets are generated at deploy time. The first organization, project, API keys and admin user are created on first boot, and sign-up is closed, so nobody else can claim a fresh URL.
Common Use Cases
- Trace and debug LLM apps and agents: prompts, tool calls, latency and errors for every request
- Track token usage and cost per model, user and feature
- Manage prompt versions and run evaluations, datasets and experiments on real traces
Dependencies for Langfuse Hosting
- Postgres, ClickHouse and Redis (included, each on its own volume)
- S3-compatible object storage (a Railway bucket, included)
Deployment Dependencies
- Langfuse on GitHub and the v4.78.0 release
- Langfuse self-hosting docs
- Start scripts used by this template
Implementation Details
Sign in: open your Railway domain and sign in with LANGFUSE_INIT_USER_EMAIL and the generated LANGFUSE_INIT_USER_PASSWORD. Both are in the Langfuse service's Variables tab.
Send traces: the project "My Project" already has API keys. Copy them from the same Variables tab into your app:
LANGFUSE_PUBLIC_KEY=pk-lf-... # value of LANGFUSE_INIT_PROJECT_PUBLIC_KEY
LANGFUSE_SECRET_KEY=sk-lf-... # value of LANGFUSE_INIT_PROJECT_SECRET_KEY
LANGFUSE_BASE_URL=https://your-app.up.railway.app # your Railway domain; LANGFUSE_HOST in older SDKs
Langfuse v4 accepts traces over OpenTelemetry, which the current Python and JavaScript SDKs use. For older SDKs, set LANGFUSE_MIGRATION_V4_WRITE_MODE=dual on both Langfuse and Langfuse-Worker.
Teammates: sign-up is off (AUTH_DISABLE_SIGNUP=true in the start script). To let invited teammates create accounts, set AUTH_DISABLE_SIGNUP=false on the Langfuse service, then set it back to true once they have joined. While it is false, anyone who can reach the URL can sign up.
Not included: e-mail (set SMTP_CONNECTION_URL and EMAIL_FROM_ADDRESS on both services for invites and password resets), SSO, and the AI Gateway (beta, needs its own domain).
Why Deploy Langfuse on Railway?
Railway is a singular platform to deploy your infrastructure stack. Railway will host your infrastructure so you don't have to deal with configuration, while allowing you to vertically and horizontally scale it.
By deploying Langfuse on Railway, you are one step closer to supporting a complete full-stack application with minimal burden. Host your servers, databases, AI agents, and more on Railway.
Template Content
Langfuse-Worker
langfuse/langfuse-worker:4.78.0Langfuse
langfuse/langfuse:4.78.0LANGFUSE_INIT_USER_EMAIL
Postgres
postgres:17.11ClickHouse
clickhouse/clickhouse-server:26.3.62.3Redis
redis:8.10.3langfuse-storage
Bucket

