Arize Phoenix
Open-source LLM tracing, evaluation, and hallucination detection.
What it does
An open-source LLM tracing, evaluation, and hallucination detection platform. Provides detailed traces of LLM interactions with built-in evaluation frameworks for measuring output quality, relevance, and factual accuracy. CB Insights positions Arize in the AI agent tech stack Oversight layer alongside Langfuse and Patronus AI. Critical for detecting ASI06 (Memory & Context Poisoning) through drift monitoring and ASI08 (Cascading Failures) through multi-agent trace analysis.
Security relevance
Hallucination detection is a security concern — LLM09 (Overreliance) becomes dangerous when models generate confident but incorrect information that users trust. Arize Phoenix's evaluation framework helps quantify hallucination rates and identify patterns in unreliable outputs.
When to use it
Use when you need to measure and monitor LLM output quality, particularly hallucination rates. Python-based with straightforward integration. Guided setup — install, instrument your application, configure evaluations.
OWASP coverage
Risks addressed — mapped to both OWASP Top 10 standards. 1 in LLM, 2 in Agentic.
The raw record
What Yuntona stores. Single source of truth — fork it on GitHub.
name: Arize Phoenix slug: arize-phoenix type: Mixed category: AI Development Tools url: https://phoenix.arize.com reviewed: 2026-04 added: 2026-04 updated: 2026-04 risks: llm: [LLM09] asi: [ASI06, ASI08] complexity: Guided Setup pricing: — audience: Builder lifecycle: [monitor] tags: [Evals, Observability, Open Source]