Phoenix
Free tierThe open-source platform for AI agent development, tracing, and evaluation
Key strengths
Facts verified against phoenix.arize.com on
Recent changes
- Starting price: $0/mo → $50/mo
- Plans updated: AX Free (free, 25k spans/mo, 1 GB, 15-day retention, SaaS); AX Pro ($50/mo, 50k spans/mo, 10 GB, 30-day retention, SaaS); AX Enterprise (custom, SaaS or Self-Hosted)
Get alerts when Phoenix changes
Free. We'll email you when pricing, the free tier, API access or open-source status changes.
Use Cases — Technical
Arize Phoenix is built for AI engineers and developers working on LLM-powered systems. Common technical use cases include:
1. LLM Application Tracing Instrument LLM calls, retrieval steps, and tool invocations with OpenTelemetry spans. Gain span-level visibility into latency, token usage, and chain-of-thought across complex pipelines.
2. AI Agent Observability Trace multi-step agent workflows — including tool calls, memory reads, and sub-agent handoffs — to debug unexpected behaviors and optimize agent performance.
3. RAG Pipeline Monitoring Monitor retrieval-augmented generation pipelines (e.g., built with LlamaIndex) to evaluate retrieval quality, context relevance, and answer faithfulness.
4. Automated Evaluation with LLM-as-Judge Run LLM-as-judge evaluations at scale over collected traces or curated datasets to score response quality, hallucination rates, and task completion — without manual review for every sample.
5. Human Annotation Workflows Combine automated scoring with structured human annotation interfaces for ground-truth labeling and quality assurance on high-stakes outputs.
6. Experiment & Prompt Management Version datasets and run controlled experiments to compare prompt variants, model versions, or retrieval configurations — enabling data-driven iteration.
7. Self-Hosted / Air-Gapped Deployments Deploy Phoenix via Docker or Kubernetes in private infrastructure environments where data cannot leave the organization's control.