Helicone's status, from its own pages
- 3 March 2026. Mintlify announced it was acquiring Helicone and said Helicone would run in maintenance mode.
- October 2026. Helicone's homepage now reads "Helicone is no longer part of Mintlify. Details to come soon."
- Today. The cloud sign-up page still says "Sign ups are disabled", while the pricing page lists Hobby (free, 10,000 requests), Pro at $79 a month and Team at $799 a month.
Nobody has published what the October change means for the product, its roadmap or your data. Treat it as a reason to plan a move calmly, not to rush one, and re-read Helicone's own pages before you decide.
Which job did Helicone do for you?
Helicone grew from a logging proxy into a gateway plus an observability platform. Most teams used a subset. Name yours first:
| If you mainly used Helicone for | Look at | Because |
|---|---|---|
| Request logs, sessions, prompt versions, evaluation | Langfuse | Open-source tracing, prompt management and evaluation, self-hostable |
| Cost and usage beside product analytics | PostHog AI Observability | LLM events in the same tool as your product data |
| A self-hosted proxy with spend tracking | LiteLLM | MIT-licensed gateway with budgets per key, team and user |
| Its gateway: one base URL, fallbacks, caching, rate limits, cost per user | A BYOK gateway such as INFRO | Hosted gateway on your own accounts, with budgets that refuse before the call |
Langfuse and PostHog record what your code sends them; they do not sit in the request path, so they cannot route or enforce a limit. A gateway can, and is one more service in front of every call.
The alternatives, side by side
Each option is described from its own pricing page and documentation, checked on the date at the top. INFRO is listed first because this is INFRO's site; its limits are listed as plainly as everyone else's.
Which one to choose
- Langfuse
- if Helicone was your tracing and evaluation tool and routing is solved elsewhere.
- PostHog AI Observability
- if you already use PostHog and want LLM cost beside product analytics.
- LiteLLM
- if you used Helicone's proxy for cost tracking and can run a gateway yourself; pair it with Langfuse for traces.
- INFRO
- if you used Helicone's gateway and want failover, budgets and cost per customer on your own provider accounts without running it.
- Portkey
- if you want a hosted gateway with logs inside a security platform.
A migration that keeps your history
- Inventory. List every service that calls Helicone, the base URL each uses (
https://ai-gateway.helicone.ai, or the olderoai.helicone.aiproxy), and everyHelicone-Property-*andHelicone-User-Idvalue you send. Those properties are your attribution model. - Export first. Pull request and cost history into your own warehouse before anything changes. Most tools cannot import another tool's traces.
- Map the fields. Each Helicone header becomes the destination's equivalent: a Langfuse trace attribute, a LiteLLM tag, or INFRO request
metadata. - Run in parallel for a week and compare daily requests and cost; a gap points at a service you missed.
- Switch one service at a time behind a flag, and keep Helicone readable until the new tool has a full month of history.
The Helicone migration guide maps every Helicone header to INFRO, with tested before-and-after code.
Questions
- Is Helicone shutting down?
- Helicone has not announced a shutdown. Its homepage says it is no longer part of Mintlify, with details to come, and its cloud sign-up page says sign-ups are disabled. Read its own pages before deciding.
- What is the best open-source alternative to Helicone?
- Langfuse for tracing, prompts and evaluation (MIT, self-hostable), and LiteLLM for a self-hosted gateway (MIT). Together they cover both halves of what Helicone did.
- Can I keep my Helicone custom properties?
- The values move, not the history. Map each Helicone-Property header to the destination's field (Langfuse trace attributes, LiteLLM tags, or INFRO request metadata) and export the old data before you switch.