Phoenix
by Arize Phoenix · Open-source tracing, evaluation and experimentation for AI agents
BenchRank score
Screenshots of Phoenix
Homepage
Overview
Phoenix is an open-source platform for developing and evaluating AI agents. It traces each step an agent takes — prompts, retrievals, tool calls and outputs — then lets you annotate runs, build datasets from traces, run experiments and score results on cost, latency and performance. It uses OpenTelemetry and runs locally, in Docker, on Kubernetes via Helm, or as a hosted Cloud instance.
- Best for
- AI engineers who need to trace, evaluate and iterate on LLM agents on their own infrastructure
- Pricing
- No prices are published for Phoenix itself — it is ELv2-licensed and self-hostable with 2 free Phoenix Cloud instances; the captured pricing page prices the separate Arize AX product instead (Free, $50/month Pro, custom Enterprise).
- Runs on
- WebSelf-hostedCLI
Strengths and trade-offs
Strengths
- Self-host so traces stay on your own infrastructure
- Native OpenTelemetry; works with any model, framework or language
- Runs locally, via Docker, on Kubernetes with Helm, or Phoenix Cloud
- Tracing, evals, datasets, experiments and a prompt IDE in one tool
Trade-offs
- Free Phoenix Cloud is capped at 2 instances
- Self-hosting means running, upgrading and scaling it yourself
- Pages steer scaled use cases to the paid Arize AX product
- No published pricing or limits for Phoenix itself
Pricing
Published plans and prices from Phoenix’s own pricing page.
How this score is made up
Each dimension is scored out of 100 and combined into the headline score using fixed weights.
MCP support
Whether an agent can drive the product through the Model Context Protocol, and how much setup that takes.
API quality
Public API surface: machine-readable spec, official SDKs, documented auth, errors, rate limits and versioning.
Documentation
Publicly reachable docs — coverage, freshness, code samples and machine readability.
Agent friendliness
How readable the site is to an automated client: llms.txt, structured data, server-rendered content, crawler access.
Pricing transparency
Whether real prices are published, self-serve signup exists, and usage costs are knowable without a sales call.
Changelog
A public, dated record of what shipped and when — the clearest signal that a product is still alive.
Marketing site structure
Whether the site answers a buyer's questions: clear positioning, the pages that matter, and accessibility.
Page speed
How fast the site loads for real visitors: Chrome UX Report 75th-percentile LCP, INP and CLS, with a Lighthouse mobile run standing in where a site has too little traffic for field data.
Operational trust
Status page and incident history, security disclosure, compliance and data-processing documentation.
Measured, but not part of the score
Useful to know, but not a mark for or against the product — so these do not affect the ranking.
Openness
Source availability, self-hosting, data export and open standards. Scored and shown, but not part of the composite — paid SaaS is not worse for being paid SaaS.
Maintenance
Release cadence and repository activity. Scored and shown, but not part of the composite — it is only measurable for open repositories.
This doesn’t look right — report a problem with Phoenix’s score
Where this comes from
The Phoenix pages BenchRank reads when it scores the product — its documentation, release notes, status and security pages, and its repository where there is one.
Alternatives in Model Hosting & Inference
Ranked 2
74.8 — BenchRank score out of 100Helicone
Helicone · AI gateway and LLM observability for routing, debugging and analysing apps
Best for: AI engineering teams routing, debugging and monitoring LLM calls across many providers
Ranked 3
73.5 — BenchRank score out of 100Langfuse
Langfuse · Open-source tracing, evaluation and prompt management for LLM apps
Best for: Engineering teams tracing, evaluating and improving LLM apps who want open source and self-hosting.
Ranked 4
73.2 — BenchRank score out of 100Replicate
Replicate · Run, fine-tune and deploy AI models through a cloud API
Best for: Developers who want to run, fine-tune or deploy AI models via an API without managing GPUs
Is this your product?
Claim Arize Phoenix to manage its profile. Claiming lets you suggest edits to the descriptive fields — it never changes scores or rankings.