Both offer a free tool you can run yourself and a hosted cloud, and both trace agents, run experiments and version prompts. Opik's free self-hosted version includes online evaluation of production traffic under an Apache-2.0 licence; Arize keeps that in AX, its paid platform, now owned by Dynatrace.[3][4]
The main difference between Arize and Opik is what the free self-hosted version includes: Opik, from Comet, is Apache-2.0 and its self-hosted release scores production traffic with online evaluation rules, while Arize's free tool, Phoenix, is source-available and points to the paid AX platform for continuous production evals. Dynatrace completed its acquisition of Arize on 1 October 2026 and says Phoenix and AX continue. Comet remains independent.
| Decision factor | Arize | Opik |
|---|---|---|
| Ownership | Acquired by Dynatrace, completed 1 October 2026; Phoenix and AX continue | Comet ML, independent; reports $70M in total funding |
| Licence | Phoenix: Elastic License 2.0 (source-available); AX: commercial | Apache-2.0 for the core platform and SDKs |
| Online evals in the free self-hosted version | No; Phoenix docs point to AX for continuous production evals | Yes, online evaluation rules on sampled traces, spans and threads |
| OpenTelemetry | Built on it; Arize maintains the OpenInference conventions | OTLP over HTTP only; gRPC not supported |
| Run-time checks | Partial: AX guards via Guardrails AI, run in your code | Partial: guardrails your code calls; they fail closed |
| Free hosted tier | AX Free: 25k spans a month, 15-day retention, unlimited users | Free Cloud: 25k spans a month, 60-day retention, up to 10 members |
| First paid tier | AX Pro: $50 a month, 50k spans, 30-day retention | Pro Cloud: $19 a month, 100k spans, then $5 per 100k |
| Capability | Arize | Opik | Prefactor |
|---|---|---|---|
| Observe | |||
| LLM call tracing | ✓ | ✓ | ✓ |
| Multi-step agent tracing (tool calls, nested spans) | ✓ | ✓ | ✓ |
| Cost tracking | ✓ | ✓ | ✓ |
| OpenTelemetry support | ✓ | ✓ | ✓ |
| Evaluate | |||
| Offline evals (datasets, experiments) | ✓ | ✓ | ✓ |
| Online evals on production traffic | ✓ | ✓ | ✓ |
| LLM-as-a-judge scoring | ✓ | ✓ | ✓ |
| Code-based evaluators | ✓ | ✓ | ✓ |
| Human review and annotation | ✓ | ✓ | ✓ |
| Prompt versioning and playground | ✓ | ✓ | — |
| Act on findings | |||
| Runtime controls on live agent actions | Partial | Partial | ✓ |
| Gateway or proxy in the request path | Partial | Partial | — |
| Licence and deployment | |||
| Open source | Partial | ✓ | Partial |
| Self-hosting | ✓ | ✓ | Partial |
Both vendors let you run a tracing and evaluation tool on your own infrastructure at no cost, but the two free versions stop in different places. Opik's self-hosted release comes from the same Apache-2.0 repository as its cloud, and Comet's documentation says it includes every Opik feature except user administration. That covers tracing, datasets and experiments, the Prompt Library, annotation queues and online evaluation rules that score sampled production traces, spans and threads.
Phoenix, Arize's self-hosted tool, covers tracing, datasets, experiments, LLM and code evaluators and prompt versioning. Continuous evaluation of production traffic, monitors, alerting and automated issue detection belong to Arize AX, and Phoenix's own documentation points users there. If scoring live traffic without a subscription is the requirement, Opik meets it and Phoenix does not.
The licences differ as well. Apache-2.0 places no limit on hosting Opik for others. Phoenix uses the Elastic License 2.0, which allows reading, modifying and self-hosting the code but forbids offering it to others as a hosted service. Most internal deployments are unaffected; a platform team building a shared service should read the terms. One operational note on Opik: its local Docker install is described as not production-ready, and production deployments run on Kubernetes.
Dynatrace completed its acquisition of Arize on 1 October 2026. Its announcement says Arize will keep supporting Phoenix and AX and that Arize capabilities will be integrated into Dynatrace over time. For a team already on Dynatrace, that integration could be the main reason to pick Arize. For everyone else, packaging and pricing may change over the next few quarters, so a large commitment is worth timing with that in mind.
Opik is built by Comet ML, founded in 2017, which reports $70M in total funding and has no ownership change on record as of 20 October 2026. Its changelog shows steady releases this year: Opik 2.0 with an environment-tagged Prompt Library in June, a per-evaluation spend cap for judges in July, and annotation queues that fill themselves from feedback-score conditions in October.
Arize maintains OpenInference, a set of OpenTelemetry conventions for LLM and agent spans, and Phoenix and AX ingest the same instrumentation, so moving between Arize's two products needs no code change. AX adds Agent Trajectory, a graph of which agents call which, set automatically for LangGraph, CrewAI, AG2 and others.
Opik accepts OTLP over HTTP only; its documentation says the gRPC exporter causes errors, and it does not state which semantic conventions it reads. Its own SDKs cover Python, TypeScript, Java and .NET, with integrations for LangGraph, CrewAI, OpenAI Agents, Google ADK, the Claude Agent SDK and many more. Teams already instrumented with OpenInference should test how Opik maps those attributes before switching.
On run-time checks the two are close. AX documents guards built on the Guardrails AI framework that run in your own code around model calls and can return a default response, re-ask or raise an exception. Opik's guardrails (PII, topic, prompt injection, judge and custom classifier checks) run as a blocking call your code must make and handle; its PII and topic checks are English only. Neither offers approval holds or policies on agent tool calls.
Both price their hosted plans on spans, with different allowances. As of 20 October 2026, AX Free includes 25k spans a month with 15-day retention and unlimited users, and AX Pro is $50 a month for 50k spans with 30-day retention. Opik's Free Cloud includes 25k spans a month for up to 10 team members with 60-day retention, and Pro Cloud is $19 a month for 100k spans and up to 50 members, then $5 per 100k spans.
Self-hosted AX is Enterprise only and priced by quote; Phoenix and open-source Opik are free to run, with the infrastructure cost on you. Check Arize's pricing and Opik's pricing against your own volumes before you decide.
We sell the layer this section describes. Read it with that in mind.
Stated plainly, since Prefactor publishes this page: Prefactor is an agent evaluation platform that scores each step of a run, not only the final answer, and can act on a failing score through runtime policies, approval holds and a kill switch for a run, an agent or a team. Arize AX and Opik both offer guardrails that your own code calls, but neither documents approval holds or policies on agent tool calls. Prefactor has no prompt playground or prompt registry and its platform is hosted, so if free self-hosting or prompt iteration is the priority, Opik or Phoenix is the better fit. See the full comparison of 15 platforms.
Reviewed against public product and documentation pages on October 20, 2026. If a vendor has changed a feature, product name, or positioning since then, send a correction and we will update it. Numbered source links in the page body point to the ordered sources below.
Prefactor scores each step of every agent run and can hold or stop an action when a score says it should not happen. Start on the free tier: 5,000 spans a month, no credit card required.
Prefactor helps teams observe, evaluate, and improve their AI agents in production, across every framework and provider.