Deploy Arize Phoenix

Open-source LLM tracing, evals and datasets, pinned with Postgres

Deploy Arize Phoenix

/var/lib/postgresql/data

Deploy and Host Arize Phoenix on Railway

Arize Phoenix is an open-source AI observability and evaluation platform. It collects OpenTelemetry traces from LLM apps and agents and shows them in a web UI. It also has evaluations, versioned datasets, experiments, a prompt playground and prompt management. It works with OpenAI, Anthropic, LangChain, LlamaIndex and other providers and frameworks.

About Hosting Arize Phoenix

This template runs Phoenix 21.1.0 from the official arizephoenix/phoenix image, pinned by digest and built from a public GitHub repo, with Railway Postgres. All Phoenix data (traces, projects, datasets, experiments, prompts, users, API keys) is stored in Postgres on a volume, so redeploys and restarts lose nothing. Login is required. The signing secret and the first admin password are generated at deploy, so you don't answer any prompts and the default admin/admin password is never accepted. The public domain serves the UI, the REST API and OTLP/HTTP trace ingestion. A health check on /readyz confirms that Phoenix can reach its database before Railway sends traffic to a new deploy.

Common Use Cases

  • Tracing LLM and agent apps (prompts, tool calls, latency, token use and cost) to debug them in development and production
  • Running LLM-as-a-judge evaluations and experiments on versioned datasets before shipping a prompt or model change
  • Managing and testing prompts in the playground, shared by a team on one self-hosted server

Dependencies for Arize Phoenix Hosting

  • Arize Phoenix server (Python, official Docker image)
  • PostgreSQL (Railway Postgres, included)

Deployment Dependencies

Implementation Details

After deploy: open the Phoenix service's URL and sign in as admin@localhost. The password is the value of PHOENIX_DEFAULT_ADMIN_INITIAL_PASSWORD in the Phoenix service's Variables tab. Change it in your profile, then create an API key under Settings → API Keys.

Send traces with the Phoenix or OpenInference libraries:

PHOENIX_COLLECTOR_ENDPOINT=https://YOUR-DOMAIN
PHOENIX_API_KEY=YOUR-API-KEY

Or send OTLP/HTTP (protobuf) to https://YOUR-DOMAIN/v1/traces. Services in the same Railway project can use OTLP/gRPC at ${{Phoenix.RAILWAY_PRIVATE_DOMAIN}}:4317.

Optional: email for invites and password resets (PHOENIX_SMTP_HOSTNAME, PHOENIX_SMTP_PORT, PHOENIX_SMTP_USERNAME, PHOENIX_SMTP_PASSWORD, PHOENIX_SMTP_MAIL_FROM), OAuth2 or LDAP sign-in, and trace retention (PHOENIX_DEFAULT_RETENTION_POLICY_DAYS). Add them as variables on the Phoenix service.

Updates: the Phoenix service builds from the template repo. When the pinned Phoenix version is bumped there, Railway shows an update notice on your project and you choose when to apply it. Phoenix migrates its database on start. Check the repo's CHANGELOG.md first.

Why Deploy Arize Phoenix 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 Arize Phoenix 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

More templates in this category

View Template
Chat Chat
Chat Chat, your own unified chat and search to AI platform.

okisdev
117
View Template
Weaviate
Vector database with hybrid search and an API key-protected API

GitHub Locker
1
View Template
Möbius Personal AI OS
Self-hosted AI agent that builds and runs apps for you, with memory.

bento
1