Launching soon — get early access:
Arize Phoenix

Arize Phoenix

Open Source(NOASSERTION)
AI-Native

The open-source platform for agent development & evaluation

SOC 2
GDPR
HIPAA
Public status page
No training on your data
+9
Last updated: Aug 28, 2026
Open Source
Visit Site11.1k
More

Quick Answer

AI

Arize Phoenix is an open-source, local-first platform for AI observability and evaluation, enabling developers to trace, evaluate, and improve LLM applications and agents. It processes 1 billion evals per year and has 5 million downloads per month. Best for AI engineers and MLOps teams needing transparent debugging and evaluation with local data control. Free to use; managed Arize AX starts from $50/month.

About Arize PhoenixAI

Arize Phoenix is an open-source platform for AI observability and evaluation, designed for developers to trace, evaluate, and improve LLM applications and agents. It integrates with popular frameworks like LangChain and LlamaIndex, offering local-first deployment for full data control. Phoenix helps debug, measure quality, and monitor performance of AI systems.

What Makes It UniqueAI

Phoenix is the leading open-source, local-first platform for AI observability and evaluation, built on OpenInference and OpenTelemetry standards, offering complete control over data and seamless integration with existing AI stacks.

AI-extracted from public sources. May contain errors. Methodology · Report an error

Decision Intelligence

Best For

AI
AI engineers needing transparent debugging and evaluation for LLM applications
Developers building and improving AI agents with open-source tools
Teams requiring local data control and custom evaluation workflows
Organizations scaling AI systems with a focus on reliability and performance
MLOps teams integrating observability into their existing data stacks

Consider Alternatives If

AI
Non-technical users seeking a no-code AI development platform
Teams requiring a fully managed, enterprise-grade MLOps solution without open-source components
Users who prefer dedicated 24/7 support without community interaction
Pricing Overview

Type

Open Source

Free Trial

No

Credit Card

Not Required

Free Plan

Yes

Dimension Scores

· 7.3 avg

How good it is, across universal quality axes

Ease of Use6/10
Output Quality8/10
Value for Money7/10
Customization8/10
Support7/10
Integration8/10
Accuracy & Reliability8/10
Compliance & Data Protection6/10
Performance8/10
AI-analyzed from product data·Medium confidenceHow we score

Task Scores

· 7.6 avg

What it can do, on category-specific tasks

Hallucination Tracking9/10
Open Telemetry Compatibility9/10
Trace Completeness8/10
Latency Percentile P95 P998/10
Retrieval Accuracy Monitoring8/10
Drift Detection Speed7/10
Token Cost Attribution7/10
Time To First Token7/10
Alert Quality7/10
Prompt Injection Detection6/10

Scores are AI-estimated from publicly available data — not an independent test or a verified user rating. How we rank →

Capabilities
Data Analysis
Code Review
ProsAI
  • +Open-source and local-first for full data control and transparency
  • +Built on open standards (OpenInference, OpenTelemetry) for broad interoperability
  • +Comprehensive evaluation framework for agents and LLM applications
  • +Integrates with a wide range of AI models and frameworks
  • +Supports self-hosting for flexible deployment
  • +Strong community support and active development
LimitationsAI
  • -Primarily focused on observability and evaluation, not a full MLOps platform
  • -Requires technical expertise for self-hosting and advanced configurations
  • -Community support is the main channel for free users, not dedicated support
  • -Limited to 25k spans/month and 1GB ingestion for the free SaaS tier (Arize AX)
  • -Phoenix is local-first, requiring manual setup for cloud deployment
Fit Score

Complexity

Advanced

Company Size

Small
Medium
Enterprise

Target Team Size

Small Team
Large Team

Target Skill Level

Expert
Technical Specifications
SDK Available
MCP Compatible
Team Features
Webhook Support
Has DemoOpen demo
Function Calling (unknown)
Sandbox Mode
Audit Logs
No Signup (unknown)
Works Offline
Requires Host App
Framework
MCP Server
Inference API
Authentication
Single Sign-On
SAML
Developer Access
API
Public
SDK
Python
Deployment:
Self Hosted
Response time (est.):
Fast
Compliance, Privacy & Trust
GDPR (as claimed): Yes
SOC 2 (as claimed): Yes
Type II
HIPAA (as claimed): Yes
Privacy & data handling
Trains on user data: No
Human review:
Recommended

Compliance audit trail

4 events
  1. GDPR

    Privacy Policy updated

    This month · 2026-08-11 · source

  2. SOC 2

    SOC 2 Type II certified

    7mo ago · 2026-01-01

  3. HIPAA

    HIPAA compliant

    7mo ago · 2026-01-01

  4. GDPR

    GDPR compliant

    7mo ago · 2026-01-01

The Model(s) Behind It

Model details — extracted from public vendor sources by our pipeline (Smart Generator + A3 cross-referencing) or admin-entered. Shown as claimed by the vendor; not independently audited by Intelloro.

Model Weight Availability
Fully open (Apache/MIT)
Training Data Provenance
Not disclosed
Relevant to EU AI Act Art. 53 / Annex XI (GPAI)
Integrations & Connections
30 connections
A
A
H
+26
Agno
Agno

Agent framework & high-performance runtime for multi-agent systems

Haystack
Haystack

Build custom AI agents and RAG applications with open AI orchestration

AU
AutogenNative

Tracing for AutoGen agent framework

GU
Guardrails AINative
HU
Hugging FaceNative
AL
AlgorithmiaNative

ML integration for Algorithmia

AN
AnthropicNative

Tracing for Anthropic LLM provider

AN
AnyscaleNative

ML integration for Anyscale

Auto-updated

Build a Stack
Video

Trust Score

88/100

Very Good

Based on 11 data signals

Confidence: Medium

Reflects publicly available data, not an independent audit

Compliance & Verified26/38
GDPRYes
SOC 2Yes
HIPAAYes
Human ReviewYes
Operational35/35
Status PageYes
Data ResidencyUS, EU, Canada
Market Proof10/10
Notable CustomersGitHub, Twilio, Shopee
Company17/17
Company ListedYes
Team Size201-500
User Base1M-10M
FundingSeries C
Health

Not yet checked

AI-Assisted Data

Some details are AI-generated estimates — verify with the vendor.

AI-generated estimate: Best For, Description, Limitations, Not Good For, Pros, Quick Answer, What Makes Unique

How we source data

AI Governance & Regulatory Disclosures

Legally-sensitive fields — extracted from public vendor sources by our pipeline (Smart Generator + A3 cross-referencing) or admin-entered. Shown as claimed by the vendor; not independently audited by Intelloro.

EU AI Act Risk Tier
Unclassified
EU AI Act Article 50 Disclosure
Not declared
COPPA Compliance (US under-13)
Not declared

Switching & Migration

Migration Difficulty
Moderate Effort
Data portable as:
CSV
JSONL
JSON

Slack + GitHub

Quick Stats
PlatformsWeb, API, CLI
Integrations30+
User Base1M-10M
LaunchedApr 2023
Company & Links
CompanyArize AI
Founded2020
Based inUnited States
Last major updateAug 2026
FundingSeries C · $131M
DocumentationChangelog
All Features
LLM call tracing
Agent evaluation framework
Production monitoring
Automated alerting for failures
Experimentation for prompt and model changes
Human annotation for output categorization
Local-first deployment
OpenInference and OpenTelemetry standards compliance
AX Free
Free
  • 25k spans per month
  • 1 GB ingestion per month
  • 15 days retention
  • SaaS deployment
  • Unlimited users
  • Unlimited evals
AX Pro
$50/month
  • 50k spans per month
  • 10 GB ingestion per month
  • 30 days retention
  • SaaS deployment
  • Unlimited users
  • Unlimited evals
  • 10 issues/month Signal
  • Email support
  • Standard SLAs
AX Enterprise
Custom
  • Custom trace spans
  • Custom ingestion volume
  • Custom retention
  • SaaS or Self-Hosted deployment
  • Unlimited users
  • Unlimited evals
  • Unlimited Signal issues
  • Unlimited Signal PRs
  • Multi-modal tracing support (image, voice, pdf)
  • Datalake sync with ADB Data Fabric
  • Agent trajectory visualizations (path and graph)
  • Custom dashboards
  • Custom metrics
  • Custom monitors
  • Custom views on traces
  • Online evals on traces/spans
  • Offline evals on datasets & experiments
  • Custom code evaluators
  • Agent as a judge
  • Align evaluators
  • Trace Evals: Evaluate agent trajectories
  • Session Evals: Evaluate multi-turn conversations
  • Multi-modal evaluation (image, voice, pdf)
  • Human annotations and feedback
  • Labeling queues
  • Playgrounds
  • Datasets
  • Experiments
  • Prompts
  • Agent Experimentation: Run your agent in playground
  • Prompt serving and versioning
  • Trace replay in playground
  • Multi-prompt comparison
  • Prompt Learning (PL) optimization
  • Orchestrate sandboxed managed debugging agents
  • Swarm observability across managed and third-party agents
  • Always-on issue detection with Signal
  • Agent as a judge
  • Self-hosted deployments
  • Enterprise SSO
  • Audit logs
  • SOC 2 Type II
  • GDPR
  • HIPAA
  • Data region: US or EU or CA
  • Organization and space RBAC
  • Service accounts
  • Dedicated troubleshooting support
  • Industry education workshops
  • Invites to exclusive event
  • Arize Customer Advisory Board
  • Custom SLAs

Pricing verified from the vendor pricing page.

User Reviews

No reviews yet. Be the first to review!

Looking for more Arize Phoenix alternatives?

Explore our complete list of alternatives with filters for free, cheaper, and higher-rated options.

View All Alternatives
Who It's For
3 roles
Software DevelopersData ScientistsComputer and Information Research Scientists
Platform, Content & Languages
1 languages
English+4 more

Accepts

Text
Code

Produces

Text
JSON
Languages: English

Frequently Asked Questions