Both trace agents, run offline and online evals and version prompts. Arize, now owned by Dynatrace, gives you a free self-hosted tool in Phoenix and continuous production evals in its paid AX platform; Maxim AI is hosted, built around pre-release simulation including voice agents, and adds blocking through its Bifrost gateway. Whether you need to self-host for free usually decides it.[3][4]
Arize offers a free self-hosted tool, Phoenix, and a paid platform, AX, where continuous production evals live; Maxim AI is a hosted platform for simulating, evaluating and tracing agents, sold alongside its open-source Bifrost gateway. Dynatrace completed its acquisition of Arize on 1 October 2026, and Arize says Phoenix and AX continue. Maxim AI is independent and is the one to look at for voice agent simulation; Arize is the one to look at for free self-hosting.
| Decision factor | Arize | Maxim AI |
|---|---|---|
| Ownership | Acquired by Dynatrace, completed 1 October 2026; Phoenix and AX continue | Independent; $3M seed led by Elevation Capital in 2024 |
| Licence | Phoenix: Elastic License 2.0 (source-available); AX: commercial | Platform proprietary; Bifrost gateway and SDKs Apache-2.0 |
| Free self-hosting | Yes, Phoenix (tracing, experiments, evaluators, prompts) | Bifrost gateway only; the evaluation platform is Enterprise in-VPC |
| Pre-release simulation | Not documented | Multi-turn text and voice agent simulation |
| Runtime guardrails | Partial: AX guards run in your code via Guardrails AI | Yes, through Bifrost Enterprise guardrails on inputs, outputs and MCP tool calls |
| OpenTelemetry | Built on it; Arize maintains OpenInference | OTLP over HTTP; GenAI, OpenInference and AI SDK conventions |
| Free hosted tier | AX Free: 25k spans a month, 15-day retention, unlimited users | Developer: up to 3 seats, 10k logs a month, 3-day retention |
| First paid tier | AX Pro: $50 a month, 50k spans | Professional: $29 per seat a month, 100k logs |
| Capability | Arize | Maxim AI | 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 | ✓ | ✓ |
| Gateway or proxy in the request path | Partial | ✓ | — |
| Licence and deployment | |||
| Open source | Partial | Partial | Partial |
| Self-hosting | ✓ | ✓ | Partial |
Dynatrace completed its acquisition of Arize on 1 October 2026. The completion announcement says Arize will keep supporting both Phoenix and AX and that Arize capabilities will be integrated into Dynatrace over time. Arize has kept shipping through the deal: in October Phoenix added code evaluators in Docker sandboxes and AX added webhooks and a renamed Decision Judge. Packaging may change as the integration proceeds, so check terms before a long contract.
Maxim AI is independent and raised a $3M seed round led by Elevation Capital in 2024. Its own direction has shifted in a different way: in October 2026 the getmaxim.ai homepage and main pricing page began leading with Bifrost, its open-source AI gateway, and the evaluation platform moved under getmaxim.ai/evals. Its public changelog shows no dated 2026 entries and the newest monthly update post covers December 2025, so recent platform changes are harder to verify than Arize's.
If you need to run the tool yourself without paying, Arize has the clearer answer. Phoenix self-hosts for free through pip, Docker or Kubernetes and includes tracing, datasets, experiments, LLM and code evaluators and prompt versioning. Its limit is that continuous evaluation of production traffic, monitors and alerting are AX features. Phoenix is also source-available under the Elastic License 2.0, which forbids offering it to others as a hosted service.
Maxim AI's evaluation platform is proprietary and self-hosts only as an Enterprise deployment in your VPC on AWS, GCP or Azure, either fully isolated or as a data plane peered to Maxim's control plane. What it offers in place of free self-hosting is a pre-release workflow: multi-turn simulation of text and voice agents, test runs over datasets for prompts, no-code agents and agents behind an HTTP endpoint, scheduled runs and a GitHub Actions integration. Arize documents datasets and experiments but no simulation of users talking to your agent.
Instrumentation is less of a deciding factor. Phoenix is built on OpenTelemetry and Arize maintains the OpenInference conventions; Maxim AI ingests OTLP over HTTP and reads OpenInference attributes, so traces from the same instrumentation can usually reach either.
Neither evaluation platform blocks live actions on its own. Arize AX documents guards built on the Guardrails AI framework that run in your code around model calls and can return a default response, re-ask or raise an exception. Arize has no first-party gateway; Phoenix added an OpenAI-compatible proxy in August 2026 that does not support tool calling and is not positioned for production traffic.
Maxim AI's blocking lives in Bifrost. Its guardrails check inputs, outputs and MCP tool-call arguments and results, and can block, redact or transform content. The pricing page lists guardrails as a Bifrost Enterprise feature, so the open-source gateway alone does not give you them. If you want checks on tool calls without writing them into your agent, Bifrost covers more of that; if you prefer the checks to sit in your own code, Arize's approach fits.
As of 20 October 2026, Arize AX Free is $0 for 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. Phoenix is free to self-host. Maxim AI's Developer tier is free for up to 3 seats and 10k logs a month with 3-day retention; Professional is $29 per seat a month with 100k logs and Business $49 per seat a month with 500k logs, with overage at $1 per 10k logs. Online evals on Maxim AI are listed from Professional upwards.
Arize charges by span volume with unlimited users; Maxim AI charges per seat with a log cap. A small team with heavy traffic tends to favour per-seat pricing, and a large team with light traffic tends to favour per-span pricing. Check Arize's pricing and Maxim AI's pricing against your own numbers.
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 and acts on what it finds through runtime policies, approval holds and a kill switch scoped to a run, an agent or a team. Of these two, Maxim AI's Bifrost guardrails are the closest to that, on Enterprise; Arize's guards run in your own code. Neither documents approval holds for agent actions. Prefactor has no prompt playground or prompt registry and its platform is hosted, so for free self-hosting Phoenix is the better pick, and for voice simulation Maxim AI is. 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.