Arize AX
Trace, evaluate, and improve your AI agents with Arize AX.
Arize AX is a managed platform for observing, evaluating and improving AI agents in production. It traces agent runs end to end, runs online and offline evaluations at span, trace and session level, and supports experiments for testing prompts and harnesses before deployment. The platform is built on the open-source OpenInference and OpenTelemetry standards, so instrumentation does not depend on a proprietary trace format.
Core capabilities include Signal, which surfaces notable issues and generates review-ready fixes; Managed Agents, which runs guided investigation and fix workflows; Swarm Observability for monitoring agent swarms; Agent-as-a-Judge evaluation; and Experiment for end-to-end change validation. The in-app AI engineering agent Alyx runs evaluations, debugs issues and proposes improvements. ADB stores traces in open formats and connects to BigQuery, Databricks or Snowflake via DataFabric. Integrations cover 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. Coding-agent workflows are supported for Cursor, Claude Code and OpenCode.
Arize AX is aimed at AI engineers and teams building and operating chatbots, RAG systems, copilots and agents. Phoenix, a separate open-source product, can be run locally or self-hosted for tracing and evaluation. The pages describe trying Arize AX and booking a demo, but do not state plan tiers, trials or pricing structure.
12 alternatives to Arize AX
Ranked by how well each tool replaces Arize AX: shared features, audience, price and popularity.
Observability built for the age of AI
Covers 2 of 15 key features and starts cheaper.
Free plan64 out of 100 match$7/moAI research and products that put safety at the frontier
Covers 0 of 15 key features and starts cheaper.
Free plan63 out of 100 match$17/moFull-stack observability for the agentic era
Covers 1 of 15 key features and starts cheaper.
Free planOpen source62 out of 100 match$19/mo- 62 out of 100 matchUsage-based
Open-Source, Unified Observability for the AI Era
Covers 1 of 15 key features and starts cheaper.
Free planOpen source60 out of 100 match$15/moOpen source platform for tracing, evaluating and improving AI agents and LLM applications.
Covers 4 of 15 key features and starts cheaper.
Free planOpen source60 out of 100 match$29/moGalileo is the AI observability and eval engineering platform where offline evals become a
Covers 1 of 15 key features.
Free plan60 out of 100 matchFreeError and performance monitoring for development teams
Covers 1 of 15 key features and starts cheaper.
Free plan59 out of 100 match$19/moOpen-source observability platform that unifies session replays, logs, traces, metrics and
Covers 1 of 15 key features and starts cheaper.
Free planOpen source58 out of 100 match$20/mo- 58 out of 100 matchUsage-based
Open-source platform to trace, evaluate, and debug AI agents
Covers 3 of 15 key features and starts cheaper.
Free planOpen source58 out of 100 match$30/moObservability for cloud-scale applications
Covers 1 of 15 key features and starts cheaper.
Free plan58 out of 100 match$8/mo