Both trace LLM calls, track cost and version prompts. Helicone has been in maintenance mode since March 2026 and is not taking new cloud sign-ups; HoneyHive is shipping, with online evaluations, annotation queues and OpenTelemetry ingest. New teams can only realistically adopt HoneyHive; Helicone users need a plan for their gateway.[3][4]
Helicone is an open-source LLM request logger and AI gateway that has been in maintenance mode since March 2026, with new cloud sign-ups disabled; HoneyHive is an active, hosted agent observability and evaluation platform built on OpenTelemetry. On 19 October 2026 Helicone said it is no longer part of Mintlify, with no new owner named. For a new deployment HoneyHive is usually the practical choice of the two, though anything beyond its free tier means talking to sales.
| Decision factor | Helicone | HoneyHive |
|---|---|---|
| Ownership | Acquired by Mintlify in March 2026; on 19 October 2026 said it is no longer part of Mintlify, no new owner named | Independent; $7.4M in total funding, last announced April 2025 |
| Status | Maintenance mode since March 2026; cloud sign-ups disabled | Active; regular changelog entries through October 2026 |
| Evaluation | Not an evaluation framework, by its own docs | Experiments, online evals, LLM and Python evaluators, annotation queues |
| Agent tracing | Header-based session grouping | Traces, agent graphs, trajectories, threads and conversation views |
| OpenTelemetry | Async SDKs built on OpenLLMetry; no generic OTLP endpoint | Accepts OTLP; supports OTel GenAI, OpenLLMetry and OpenInference |
| Gateway | AI Gateway for 100+ providers with failover and caching | None; integrates with LiteLLM, Portkey, Cloudflare and Vercel gateways |
| Self-hosting | Apache-2.0; Docker, Compose or Helm | Enterprise only, documented on AWS |
| Paid plans | Pro $79 and Team $799 a month, existing accounts only | No published self-serve paid tier; Enterprise by quote |
| Capability | Helicone | HoneyHive | 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 acquired Helicone on 3 March 2026 and moved it to a maintenance-only phase: patches, bug fixes and new model support, plus help for customers moving to other platforms. Cloud sign-ups were disabled the same day. On 19 October 2026 Helicone's site began saying it is no longer part of Mintlify, with details to come. Its self-hosted code remains Apache-2.0, but its separate Rust gateway, licensed GPL-3.0, was last updated in November 2025.
HoneyHive is independent and small: its last announced funding was $7.4M in April 2025, and third-party directories list about 18 to 19 staff. Its changelog shows steady releases, including a conversation view for traces in September 2026 and Salesforce Agentforce and Databricks Agent Bricks tracing in August. Weigh company size when you plan a long commitment, but between these two only HoneyHive is accepting new customers.
In May 2026 Helicone's dashboard and log processing were down for about four days after AWS locked its account; the proxy stayed up and queued logs were processed afterwards.
Helicone logs model requests and their cost, through a proxy or async SDKs, and groups them into sessions with headers rather than a span model for agent steps. Its prompt versioning deploys prompts by ID through its gateway. Its docs say it does not run evaluations, and it turned off Python evaluators in October 2026.
HoneyHive traces agent runs as spans, with agent graphs and trajectories, and evaluates them. Experiments run against datasets with CI regression gating through GitHub Actions; online evaluations score sampled production traces with LLM, Python and human evaluators; annotation queues route flagged traces to experts. It also has prompt versioning, a playground and alerts. Its full tracing SDK is Python; the TypeScript package is a REST client, and OTLP covers other languages.
Cost tracking overlaps. Helicone gives exact per-request cost through its gateway's model registry and alerts on cost thresholds. HoneyHive has cost, token and latency dashboards and alerts, session-level cost and, since July 2026, usage reports by workspace and project.
HoneyHive also traces coding agents such as Claude Code, Codex CLI and Devin. In June 2026 Microsoft named it the observability partner in its Agent Control Specification announcement, in which HoneyHive observes and the specification enforces.
On Enterprise, HoneyHive Policies set organisation-wide evaluation and monitoring standards; they do not enforce anything at run time.
Instrumentation is the easier part. Helicone's async logging uses OpenLLMetry, and HoneyHive accepts OTLP with OpenLLMetry, OTel GenAI and OpenInference conventions and an OpenTelemetry Collector integration. Teams logging through Helicone's proxy instead will need to add HoneyHive's SDK or an OpenTelemetry instrumentor.
The gateway is the harder part. HoneyHive has no gateway for application traffic; it traces through third-party gateways such as Cloudflare AI Gateway, LiteLLM, Portkey, TrueFoundry and Vercel AI Gateway. If you route production traffic through Helicone for failover or caching, choose one of those, or keep self-hosted Helicone running with its maintenance-only updates while you test the change.
Prompts versioned in Helicone need recreating in HoneyHive, where deployed prompts are fetched by name or version from the Python SDK or the TypeScript API client. Self-hosting changes too: Helicone self-hosts fully under Apache-2.0, while HoneyHive's self-hosted or hybrid deployment is Enterprise only and documented on AWS EKS.
As of 20 October 2026, HoneyHive's Developer tier is free with 10,000 events a month, 30-day retention and up to five users. There is no published self-serve paid plan, so any volume above that, and self-hosting, means an Enterprise quote. Helicone lists Hobby free with 10K requests and 7-day retention, Pro at $79 a month and Team at $799 a month, for existing accounts only. Check Helicone's pricing and HoneyHive's pricing before you plan a budget.
HoneyHive's free tier runs as multi-tenant SaaS in AWS us-west-2.
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 on tool calls and data access, approval holds for risky actions, and a kill switch for a run, an agent or a team. HoneyHive observes and scores but does not block or hold live actions, and Helicone's gateway can block prompt injection for OpenAI models only. Prefactor has no prompt playground or prompt registry, no gateway, and its platform is hosted, so it does not replace Helicone's routing. 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.