Helicone logs LLM requests through a proxy and Maxim AI evaluates agents, and both sell a gateway. Helicone has been in maintenance mode since March 2026 and new cloud accounts cannot be opened, so for a new project Maxim AI is the one you can actually start on. For existing Helicone users the real question is where to migrate and when.[3][4]
Helicone is an open-source LLM request logger and gateway that has been in maintenance mode since March 2026, with new cloud sign-ups disabled; Maxim AI is an actively sold platform for simulating, evaluating and tracing agents, with its own open-source gateway, Bifrost. On 19 October 2026 Helicone said it is no longer part of Mintlify, with no new owner named. For a new deployment Maxim AI is usually the practical choice; existing Helicone users should plan a migration.
| Decision factor | Helicone | Maxim AI |
|---|---|---|
| Status | Maintenance mode since March 2026; cloud sign-ups disabled | Actively sold; Bifrost v2.3.1 released 19 October 2026 |
| Ownership | Unclear: acquired by Mintlify in March 2026; says it left Mintlify on 19 October 2026, no new owner named | Independent; $3M seed led by Elevation Capital in 2024 |
| Licence | Apache-2.0 platform; separate Rust gateway GPL-3.0, last updated November 2025 | Platform proprietary; Bifrost gateway and SDKs Apache-2.0 |
| Agent tracing | Header-based session grouping, no span model for agent steps | Traces, spans, tool calls and sessions, with agent evaluators |
| Evaluation | Not an evaluation framework by its own docs; Python evaluators turned off October 2026 | Offline test runs, simulation, online evals, human review |
| OpenTelemetry | OpenLLMetry-based SDKs; no generic OTLP endpoint | OTLP over HTTP ingestion |
| Self-hosting | Free via Docker or Kubernetes; on-prem on Enterprise | Bifrost free; evaluation platform Enterprise in-VPC |
| Capability | Helicone | Maxim AI | Prefactor |
|---|---|---|---|
| Observe | |||
| LLM call tracing | ✓ | ✓ | ✓ |
| Multi-step agent tracing (tool calls, nested spans) | Partial | ✓ | ✓ |
| Cost tracking | ✓ | ✓ | ✓ |
| OpenTelemetry support | Partial | ✓ | ✓ |
| Evaluate | |||
| Offline evals (datasets, experiments) | Partial | ✓ | ✓ |
| Online evals on production traffic | Partial | ✓ | ✓ |
| LLM-as-a-judge scoring | Partial | ✓ | ✓ |
| Code-based evaluators | — | ✓ | ✓ |
| Human review and annotation | Partial | ✓ | ✓ |
| Prompt versioning and playground | ✓ | ✓ | — |
| Act on findings | |||
| Runtime controls on live agent actions | Partial | ✓ | ✓ |
| Gateway or proxy in the request path | ✓ | ✓ | — |
| Licence and deployment | |||
| Open source | ✓ | Partial | Partial |
| Self-hosting | ✓ | ✓ | Partial |
Mintlify announced its acquisition of Helicone on 3 March 2026 and said the product would move to a maintenance-only phase: patches, bug fixes and new model support, with help for customers migrating to other platforms. The cloud sign-up page has shown that sign-ups are disabled since the same day. Helicone's changelog has no feature entries after November 2025.
On 19 October 2026 Helicone replaced its acquisition messaging with a notice that it is no longer part of Mintlify, with details to come. No new owner or structure is named. In the same week it shipped hardening fixes for self-hosted deployments, turned off Python evaluators and rate limited LLM-as-a-judge calls. Earlier, in May 2026, a four-day lock on its AWS account took the dashboard and log processing down while the proxy stayed up.
Maxim AI is independent and selling. Its homepage changed in October 2026 to lead with Bifrost, and Bifrost release v2.3.1 was published on 19 October 2026. Its evaluation platform's public changelog shows no dated 2026 entries, which is worth asking about in a trial, but new accounts can be opened on every plan.
Helicone offers three main things: the proxy or AI Gateway in the request path, request logs with cost tracking, and prompt versioning deployed by prompt ID. Maxim AI maps onto each. Bifrost is an OpenAI-compatible gateway with provider failover, load balancing, semantic caching and virtual keys, so the request path moves to Bifrost. Maxim AI's platform logs token usage, latency and cost per request with threshold alerts, and its prompt tooling has versions and deployment rules that release changes without code changes.
What changes is the data model. Helicone groups requests into sessions using headers, without spans for agent steps. Maxim AI traces sessions, spans, generations, retrievals and tool calls, and ingests OTLP over HTTP. Framework coverage overlaps: Helicone lists the OpenAI Agents SDK, LangGraph, CrewAI and LlamaIndex, and Maxim AI lists the same four among more than twenty integrations. Plan to re-instrument rather than translate headers, and export historical logs before you switch, since Maxim AI's free tier keeps logs for 3 days and Professional for 7.
Self-hosted Helicone remains Apache-2.0 and will keep running, but with maintenance-only updates and no clear owner. Teams with no evaluation needs can stay longer; teams that need evals have less reason to, since Helicone describes itself as not an evaluation framework.
Helicone's prompt-threat detection can block prompt injection and jailbreak attempts, with OpenAI models only, and it supports custom rate limits. Bifrost's guardrails check inputs, outputs and MCP tool-call arguments and results and can block, redact or transform content, and Bifrost adds budgets per virtual key, team and customer. Guardrails are a Bifrost Enterprise feature. Neither documents approvals or policies for agent tool actions.
Cost control works differently. Helicone prices requests from a registry of more than 300 models, slices cost by custom properties and sends graduated cost alerts, but describes no hard spend caps. Maxim AI logs cost per request with threshold alerts and custom token prices, and Bifrost adds budgets and rate limits per virtual key, team and customer in the request path.
Helicone still lists its plans as of 20 October 2026: Hobby free with 10K requests and 7-day retention, Pro at $79 a month, Team at $799 a month and Enterprise custom. Because cloud sign-ups are disabled, new customers cannot start on any of them. Maxim AI's Developer tier is free for up to 3 seats and 10k logs a month; 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. Bifrost's open-source edition is free and Bifrost Enterprise is custom. See Helicone's pricing and Maxim AI's pricing.
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. Helicone blocks prompt injection for OpenAI models only, and Maxim AI blocks content through Bifrost Enterprise guardrails; neither documents approval holds. For Helicone users moving on, Prefactor is one destination for evaluation, but it has no gateway, prompt playground or prompt registry and its platform is hosted, so it does not replace Helicone's proxy the way Bifrost does. 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.