Helicone is an open-source LLM observability platform with gateway features — request logging, tracing, cost attribution, and proxying. This search exists because of a specific event: Mintlify acquired Helicone in March 2026, and the platform is now in maintenance mode — security updates, bug fixes and new model support continue, while active feature development has ended. That is the fact that should drive your decision, and it is stated plainly in Helicone's own announcement while being almost absent from the vendor roundups ranking on this query.
Bias declared, and it cuts against us here: Kunavo does not do observability at all — no logging, no tracing, no evals. For the monitoring half of what Helicone did we are not an option and this page will not pretend otherwise. We are relevant only if you used Helicone as a gateway, and even then it is a different arrangement.
First: which half did you actually use?
"Helicone alternatives" returns two incompatible kinds of answer because Helicone did two things. Splitting them eliminates most of the list:
- Observability — you added Helicone to see what your LLM calls were doing: request logs, latency, token counts, cost per user or feature, evals. The replacement is another observability tool, and the gateway section below is irrelevant to you.
- Gateway — you routed calls through it: one endpoint, key management, failover across providers. The replacement is a gateway, and observability tools will not cover it.
- Both — then you are replacing two things, and doing it with two tools is usually cleaner than finding one product that does both adequately.
If you used it for observability
The open-source options are where most teams leaving Helicone land, because they preserve the property that made Helicone attractive in the first place — you can self-host and keep request bodies inside your own infrastructure. Langfuse is the most common destination and covers tracing, evals and cost attribution; Arize Phoenix comes from the ML-observability side and is strong on evaluation. Both are self-hostable, and both are named here as category pointers — we have not run a comparison of them, and a gateway vendor ranking observability tools would not be worth much if we had.
Several hosted platforms also occupy this search and rank themselves first on it. That is not a reason to discount them, but it is a reason to read their comparisons knowing who wrote them — including this one.
If you used it as a gateway
Then the question is the same one every gateway choice starts with: who holds the provider credentials? Helicone is BYO-key — your accounts, its proxy. The like-for-like replacements keep that arrangement: LiteLLM self-hosted, or a hosted BYO-key control plane. Note that the most-cited hosted option, Portkey, is itself being acquired by Palo Alto Networks — moving from one acquired vendor to another is a specific choice, so make it deliberately.
Kunavo is the other arrangement: we hold the upstream credentials, so there are no provider accounts to manage and no way to route your own. It suits teams for whom holding the accounts was itself the burden, and it is the wrong answer if you have negotiated provider rates. Kunavo vs Helicone · the four gateway categories
The case for staying, which most of these lists skip
Maintenance mode is not shutdown. The service runs, it gets security patches and new model support, and the source is on GitHub, so a self-hosted exit exists no matter what happens to the hosted product. If your deployment is stable and you were not waiting on a roadmap item, migrating costs you real engineering time to solve a problem you may not have. The honest trigger for moving is narrower than "they got acquired": you need a feature that will now never ship, your contract is up for a multi-year renewal, or the hosted terms change.
What this says about picking vendors in this category
Helicone was acquired in March 2026. Palo Alto Networks announced its intent to acquire Portkey in April 2026. Two independent vendors in this category consolidated into larger platforms inside two months, which makes "will this still be independent in two years" a real selection criterion rather than a talking point.
The useful hedge is not picking a vendor nobody would buy — that is not predictable, and this page is written by a vendor, so treat any claim of permanence here with the suspicion it deserves. The hedge is preferring tools with a credible exit: an open-source version you could self-host, or a wire protocol standard enough that leaving is a base URL change. Helicone had the first, which is exactly why its acquisition is survivable. Most gateways have the second — every product here speaks either the OpenAI or the Anthropic wire, so switching costs are far lower than a feature comparison implies.
FAQ
What happened to Helicone?
Mintlify acquired Helicone in March 2026. Per Helicone's own announcement and Mintlify's, founders Justin Torre and Cole Gottdank joined Mintlify in San Francisco, and the platform — which had served around 16,000 organizations — continues in maintenance mode: security updates, bug fixes and new model support, with active feature development ended. It is still open source, so self-hosting remains available regardless of the hosted product's roadmap.
What is the best Helicone alternative?
For LLM observability specifically — logging, tracing, cost attribution, evals — the open-source options most teams move to are Langfuse and Arize Phoenix, both self-hostable. Several hosted platforms also occupy this search. If you used Helicone mainly as a gateway rather than as monitoring, the replacement is a gateway, not an observability tool, and that is a different list.
Is Helicone still usable after the acquisition?
Yes. Maintenance mode means the service runs and gets security patches, bug fixes and support for new models — it does not mean shutdown. If your deployment is stable and you were not waiting on a roadmap item, staying is a reasonable decision, and the open-source repository means you retain a self-hosted exit even if the hosted product changes. What you should not do is start a new multi-year commitment on it without pricing in that feature work has stopped.
Is Kunavo a Helicone alternative?
Not for observability — Kunavo does not do logging, tracing or evals, and there is no version of this page where we are the answer to "replace my LLM monitoring". Kunavo is relevant only to the gateway half of what Helicone did, and even there it is a different arrangement: Helicone proxies requests through provider accounts you hold, while Kunavo holds the upstream credentials and bills you for usage.
What is the difference between an LLM observability tool and an LLM gateway?
An observability tool records what happened — requests, responses, latency, token counts, cost per user or feature — and is something you can add without changing how calls are routed. A gateway sits in the request path and decides where the call goes, handling auth, routing, failover and spend limits. Several products do both, which is why the categories are often confused, and why "Helicone alternatives" returns two different kinds of answer. Work out which half you actually used before picking a replacement.
Should the acquisition change how I pick a gateway vendor?
It is fair evidence that vendor independence is a real selection criterion in this category rather than a talking point. Helicone was acquired by Mintlify in March 2026, and Palo Alto Networks announced its intent to acquire Portkey in April 2026. Two consolidations in two months is a pattern. The practical hedge is not to pick a vendor nobody will ever buy — it is to prefer tools with a credible exit: an open-source version you could self-host, or a wire protocol standard enough that switching is a base URL change.