Phoenix
Officialby
Open-source AI observability platform enabling tracing, evaluation, dataset versioning, experiment tracking, prompt management, and interactive playground for LLM applications.
Phoenix Overview
What is Phoenix about?
Provides end‑to‑end observability for large language model (LLM) workflows: captures runtime traces via OpenTelemetry, runs automated evaluations, stores versioned datasets, tracks experiments, and offers a UI playground for prompt engineering.
How to use Phoenix?
- Installation – install the server with Python or Docker.
Or pull the container:pip install arize-phoenixdocker pull arizephoenix/phoenix. - Start the server – run the Docker image or launch the Python package (
arize-phoenix serve). - Instrument your code – add the appropriate OpenTelemetry wrapper (e.g.,
openinference-instrumentation-openai,openinference-instrumentation-langchain, etc.) or use the providedarize-phoenix-otelpackage for a simplified setup. - Interact via UI – open
http://localhost:8080(default) to explore traces, evaluation results, datasets, and experiments. - Programmatic access – use
arize-phoenix-client(Python) or@arizeai/phoenix-client(JS/TS) to create datasets, upload evaluations, and query metadata.
Key Features
- Tracing – OpenTelemetry‑based capture of LLM calls across frameworks (LangChain, LlamaIndex, Haystack, etc.).
- Evaluation – built‑in response and retrieval evals; customizable eval pipelines via
arize-phoenix-evals. - Datasets & Experiments – versioned example sets, experiment tracking for prompt/LLM changes.
- Playground & Prompt Management – UI for prompt iteration, model comparison, replay of traced calls, and version‑controlled prompt assets.
- Vendor & Language Agnostic – supports OpenAI, Anthropic, Bedrock, Vertex AI, MistralAI, LiteLLM, Google GenAI, and more.
- Deploy Anywhere – local, Jupyter, container, Kubernetes, or managed cloud instance.
- Extensible SDKs – Python sub‑packages and TypeScript packages for client integration and evaluation.
Use Cases
- RAG pipeline debugging – trace retrieval and generation steps to pinpoint latency or relevance issues.
- Prompt engineering – iteratively test prompts in the Playground, compare model outputs, and roll back to prior versions.
- Model benchmarking – run standardized evals across multiple LLM providers to select the best performer.
- Continuous monitoring – integrate with CI/CD to automatically capture traces and evaluation metrics for every deployment.
- Agent observability – monitor multi‑agent workflows (e.g., CrewAI, Smolagents) for state transitions and decision points.
FAQ
Q: Do I need a cloud account to run Phoenix? A: No. Phoenix can run locally via Docker or the Python package. Cloud‑hosted instances are optional.
Q: Which frameworks are supported out of the box? A: LangChain, LlamaIndex, Haystack, DSPy, Smolagents, OpenAI SDK, Bedrock, Vertex AI, MistralAI, LiteLLM, and many more via OpenInference.
Q: How are evaluations stored? A: Evaluations are persisted as part of the Phoenix backend and can be queried through the UI or client SDKs.
Q: Can I extend the evaluation suite?
A: Yes. Custom eval functions can be added using arize-phoenix-evals or the TypeScript counterpart.
Q: Is there a licensing cost? A: Phoenix is released under the Elastic License 2.0 and is free to use; commercial support is offered by Arize AI.
Phoenix's README
Phoenix is an open-source AI observability platform designed for experimentation, evaluation, and troubleshooting. It provides:
- Tracing - Trace your LLM application's runtime using OpenTelemetry-based instrumentation.
- Evaluation - Leverage LLMs to benchmark your application's performance using response and retrieval evals.
- Datasets - Create versioned datasets of examples for experimentation, evaluation, and fine-tuning.
- Experiments - Track and evaluate changes to prompts, LLMs, and retrieval.
- Playground- Optimize prompts, compare models, adjust parameters, and replay traced LLM calls.
- Prompt Management- Manage and test prompt changes systematically using version control, tagging, and experimentation.
Phoenix is vendor and language agnostic with out-of-the-box support for popular frameworks (🦙LlamaIndex, 🦜⛓LangChain, Haystack, 🧩DSPy, 🤗smolagents) and LLM providers (OpenAI, Bedrock, MistralAI, VertexAI, LiteLLM, Google GenAI and more). For details on auto-instrumentation, check out the OpenInference project.
Phoenix runs practically anywhere, including your local machine, a Jupyter notebook, a containerized deployment, or in the cloud.
Installation
Install Phoenix via pip or conda
pip install arize-phoenix
Phoenix container images are available via Docker Hub and can be deployed using Docker or Kubernetes. Arize AI also provides cloud instances at app.phoenix.arize.com.
Packages
The arize-phoenix package includes the entire Phoenix platfom. However if you have deployed the Phoenix platform, there are light-weight Python sub-packages and TypeScript packages that can be used in conjunction with the platfrom.
Subpackages
| Package | Language | Description |
|---|---|---|
| arize-phoenix-otel | Python  |
Provides a lightweight wrapper around OpenTelemetry primitives with Phoenix-aware defaults |
| arize-phoenix-client | Python  |
Lightweight client for interacting with the Phoenix server via its OpenAPI REST interface |
| arize-phoenix-evals | Python  |
Tooling to evaluate LLM applications including RAG relevance, answer relevance, and more |
| @arizeai/phoenix-client | JavaScript  |
Client for the Arize Phoenix API |
| @arizeai/phoenix-evals | TypeScript  |
TypeScript evaluation library for LLM applications (alpha release) |
| @arizeai/phoenix-mcp | JavaScript  |
MCP server implementation for Arize Phoenix providing unified interface to Phoenix's capabilities |
Tracing Integrations
Phoenix is built on top of OpenTelemetry and is vendor, language, and framework agnostic. For details about tracing integrations and example applications, see the OpenInference project.
Python Integrations
| Integration | Package | Version Badge |
|---|---|---|
| OpenAI | openinference-instrumentation-openai |
 |
| OpenAI Agents | openinference-instrumentation-openai-agents |
 |
| LlamaIndex | openinference-instrumentation-llama-index |
 |
| DSPy | openinference-instrumentation-dspy |
 |
| AWS Bedrock | openinference-instrumentation-bedrock |
 |
| LangChain | openinference-instrumentation-langchain |
 |
| MistralAI | openinference-instrumentation-mistralai |
 |
| Google GenAI | openinference-instrumentation-google-genai |
 |
| Google ADK | openinference-instrumentation-google-adk |
 |
| Guardrails | openinference-instrumentation-guardrails |
 |
| VertexAI | openinference-instrumentation-vertexai |
 |
| CrewAI | openinference-instrumentation-crewai |
 |
| Haystack | openinference-instrumentation-haystack |
 |
| LiteLLM | openinference-instrumentation-litellm |
 |
| Groq | openinference-instrumentation-groq |
 |
| Instructor | openinference-instrumentation-instructor |
 |
| Anthropic | openinference-instrumentation-anthropic |
 |
| Smolagents | openinference-instrumentation-smolagents |
 |
| Agno | openinference-instrumentation-agno |
 |
| MCP | openinference-instrumentation-mcp |
 |
| Pydantic AI | openinference-instrumentation-pydantic-ai |
 |
| Autogen AgentChat | openinference-instrumentation-autogen-agentchat |
 |
| Portkey | openinference-instrumentation-portkey |
 |
JavaScript Integrations
| Integration | Package | Version Badge |
|---|---|---|
| OpenAI | @arizeai/openinference-instrumentation-openai |
 |
| LangChain.js | @arizeai/openinference-instrumentation-langchain |
 |
| Vercel AI SDK | @arizeai/openinference-vercel |
 |
| BeeAI | @arizeai/openinference-instrumentation-beeai |
|
| Mastra | @arizeai/openinference-mastra |
 |
Java Integrations
| Integration | Package | Version Badge |
|---|---|---|
| LangChain4j | openinference-instrumentation-langchain4j |
 |
| SpringAI | openinference-instrumentation-springAI |
 |
Platforms
| Platform | Description | Docs |
|---|---|---|
| BeeAI | AI agent framework with built-in observability | Integration Guide |
| Dify | Open-source LLM app development platform | Integration Guide |
| Envoy AI Gateway | AI Gateway built on Envoy Proxy for AI workloads | Integration Guide |
| LangFlow | Visual framework for building multi-agent and RAG applications | Integration Guide |
| LiteLLM Proxy | Proxy server for LLMs | Integration Guide |
Community
Join our community to connect with thousands of AI builders.
- 🌍 Join our Slack community.
- 📚 Read our documentation.
- 💡 Ask questions and provide feedback in the #phoenix-support channel.
- 🌟 Leave a star on our GitHub.
- 🐞 Report bugs with GitHub Issues.
- 𝕏 Follow us on 𝕏.
- 🗺️ Check out our roadmap to see where we're heading next.
- 🧑🏫 Deep dive into everything Agents and LLM Evaluations on Arize's Learning Hubs.
Breaking Changes
See the migration guide for a list of breaking changes.
Copyright, Patent, and License
Copyright 2025 Arize AI, Inc. All Rights Reserved.
Portions of this code are patent protected by one or more U.S. Patents. See the IP_NOTICE.
This software is licensed under the terms of the Elastic License 2.0 (ELv2). See LICENSE.
Phoenix Reviews
Login Required
Please log in to share your review and rating for this MCP.
Actions
Phoenix's Information
- Author
- Arize-ai
- Category
- Monitoring & Observability
- Language
- Jupyter Notebook
- License
- Other
- Stars
- 6,707
- Updated
- 2 months ago
Configuration
{
"mcpServers": {
"phoenix-mcp": {
"command": "npx",
"args": [
"-y",
"@arizeai/phoenix-mcp"
],
"env": {
"API_KEY": "<YOUR_API_KEY>"
}
}
}
}Claude Code (Terminal)
claude mcp add phoenix-mcp npx -y @arizeai/phoenix-mcpConfigure Clients
Similar MCP Servers like Phoenix
Explore related MCPs that share similar capabilities and solve comparable challenges
Netdata
by netdata
Delivers real‑time, per‑second infrastructure monitoring with zero‑configuration agents, on‑edge machine‑learning anomaly detection, and built‑in dashboards.
Tianji
by msgbyte
Provides integrated website traffic analysis, uptime checking, and server health monitoring in a single self‑hosted platform.
Grafana MCP Server
by grafana
Provides programmatic access to a Grafana instance and its surrounding ecosystem through the Model Context Protocol, enabling AI assistants and other clients to query and manipulate dashboards, datasources, alerts, incidents, on‑call schedules, and more.
Dynatrace MCP Server
by dynatrace-oss
Provides a local server that enables real‑time interaction with the Dynatrace observability platform, exposing tools for querying data, retrieving problems, sending Slack notifications, and integrating AI assistance.
Logfire MCP Server
by pydantic
Provides tools to retrieve and query OpenTelemetry trace and metric data from Pydantic Logfire, allowing LLMs to analyze distributed traces and run arbitrary SQL queries against telemetry records.
VictoriaMetrics MCP Server
by VictoriaMetrics-Community
Provides a Model Context Protocol server exposing read‑only VictoriaMetrics APIs, enabling seamless monitoring, observability, and automation through AI‑driven assistants.
Datadog MCP Server
by GeLi2001
Enables interaction with the Datadog API through a Model Context Protocol server, providing access to monitors, dashboards, metrics, logs, events, and incident data.
Last9 MCP Server
by last9
Provides AI agents with real‑time production context—including logs, metrics, traces, and alerts—through a Model Context Protocol server, enabling automatic code fixing and faster debugging.
Metoro MCP Server
by metoro-io
Provides an MCP server that exposes Metoro's eBPF‑based telemetry APIs to large language models, enabling AI‑driven queries and insights about Kubernetes clusters.