Arize is an AI observability and evaluation platform that helps engineering teams trace, debug, evaluate, and continuously improve AI agents and LLM applications in production.
About
Arize is an AI engineering platform built for teams that need to understand, measure, and improve AI agents and LLM applications once those systems reach production. Debugging an agent is difficult because failures span many steps — tool calls, prompts, retrievers, and model responses — and Arize addresses this by collecting traces, running evaluations, and turning production signals into feedback that teams can act on.
The platform is organized around three workflows: observing what an agent actually did during a request, evaluating whether agent quality is improving or regressing, and testing proposed fixes before deployment. Arize offers Arize AX, a managed SaaS platform, alongside Phoenix, an open-source observability and evaluation tool that developers can run locally or self-host. Both are built on the OpenInference instrumentation standard, which is based on OpenTelemetry. The company, Arize AI, also maintains documentation, guides, and a newsletter called The Evaluator for AI engineering practitioners.
Key Features
Pricing & Fees
- Pricing: Arize AX has three plans — AX Free at $0 (25k spans/month, 1 GB ingestion, 15-day retention, 10 Signal issues/month), AX Pro at $50 per month (50k spans/month, 10 GB ingestion, 30-day retention, 25 Signal issues/month), and AX Enterprise with custom limits and SaaS or self-hosted deployment. Startup pricing is available.
- Free tier: AX Free requires no credit card, includes unlimited users, unlimited evals, and unlimited experiments.
- Integrations: Arize works with 40+ models, frameworks, and tools including OpenAI, Anthropic, Google, Amazon Bedrock, LangGraph, LangChain, LlamaIndex, CrewAI, OpenAI Agents SDK, and DSPy, across Google Cloud, AWS, Azure, and self-hosted environments.
- Security and compliance: Arize holds SOC 2 Type II, ISO 27001, PCI DSS, HIPAA, and GDPR certifications, with flexible deployment options.
- Open source: Phoenix is available as an open-source alternative for teams that need local or self-hosted tracing and evaluation.
Who It's For
Arize is designed for AI and ML engineers, platform teams, and engineering leads who build and operate production AI systems such as chatbots, RAG pipelines, copilots, and autonomous agents. It suits organizations that need to debug agent failures, measure quality with evaluations, monitor production performance, and iterate on prompts and workflows. Teams with strict compliance requirements can use Arize AX Enterprise with self-hosting, while individual builders and smaller teams can start with the free tier or the open-source Phoenix offering.
FAQ
What is Arize?Arize is an AI observability and evaluation platform that helps teams trace, debug, evaluate, and improve AI agents and LLM applications, available as the managed Arize AX platform and the open-source Phoenix tool.
Is there a free plan for Arize?Yes, AX Free costs $0 with no credit card required and includes 25k trace spans and 1 GB of ingestion per month, 15-day retention, and unlimited users and evaluations; Phoenix is also free and open source.
Does Arize work with common AI frameworks?Arize integrates with more than 40 models, frameworks, and tools, including OpenAI, Anthropic, Amazon Bedrock, LangGraph, LangChain, LlamaIndex, and CrewAI, and runs across Google Cloud, AWS, Azure, and self-hosted environments.
What is the difference between Arize AX and Phoenix?Phoenix is the open-source observability and evaluation platform that developers can run locally or self-host, while Arize AX is the managed platform built on the same open standards with additional infrastructure, online evals, compliance certifications, and continual-improvement workflows.
How does Arize handle data privacy?Arize states that customer data stays under the customer's control, supports flexible deployment options, and is certified to SOC 2 Type II, ISO 27001, PCI DSS, HIPAA, and GDPR standards.
Related Tutorials
View all →
Related Tools
View all →
PentAGI - Advanced AI-Powered Penetration Testing Fully autonomous AI Agent for complicated penetration testing tasks. 
Orca Orca is a free, open-source Agent Development Environment that runs CLI coding agents like Claude Code, Codex, and Gemini in parallel across isolated git worktrees. 
Keygraph - Continuous Agentic Pentesting for the Enterprise Continuous agentic pentesting and AppSec for the enterprise. Autonomous exploits validated against your live app, inside your perimeter. Built by the Shannon team. 
Cursor Cursor is an AI coding agent and development environment that turns ideas into working code, assisting software engineers and teams across desktop, terminal, and cloud workflows. 
Kilo Kilo Code is an open source AI coding agent for VS Code, JetBrains IDEs, the CLI, and the cloud, offering 500+ models with zero markup and local model support.