Guides / culpa vs helicone
Culpa vs Helicone
Helicone is an open-source AI gateway and observability platform, acquired by Mintlify and described by its own team as running in maintenance mode. It tracks cost per session and per user on its own cloud. Culpa, a local-first LLM cost, margin, and forecast ledger, keeps that on your infrastructure and adds margin against revenue and a forecast.
Why this happens
Start with the fact that changes the decision. Helicone announced on 2026-03-03 that it had been acquired by Mintlify, and its own post says services stay live in maintenance mode, with security updates, new models and bug fixes still shipping. That's a fair description of a supported product rather than a dead one, and it's also a reason to ask where the roadmap goes before you build a cost practice on it. The deeper difference has nothing to do with the acquisition, and it isn't attribution either. Helicone's own pricing table lists Sessions, User analytics and Custom properties on every tier including the free one, so it does group cost by conversation and by user. What it doesn't publish is margin against what a customer pays you, a forecast of next month from your own history, or a design where the prompts stay on your infrastructure rather than on a vendor's with a retention tier attached.
What this usually looks like
- You have request-level logs and still can't name the conversation that drove last month's spike.
- Your observability retention is a plan tier rather than a decision you made.
- Nobody can tell you whether a specific customer is profitable at their current usage.
- You know the total bill and you can't forecast next month's from your own history.
Free, no card, no account
Run the free Cost Leak Scan
It shows your most expensive conversation before you install anything.
Mistakes that cost the most
| Mistake | Why it hurts | Do instead |
|---|---|---|
| Reading maintenance mode as either dead or fine. | It's neither, and treating it as either skips the question of where the roadmap goes. | Take the announcement at its word, then decide how much new capability you need from this layer. |
| Assuming cost attribution and margin are the same feature. | Attribution tells you a customer cost $80. Margin needs that set against what they pay you. | Check whether your tool holds revenue at all. Most observability tools deliberately don't. |
| Letting a plan tier decide how long your prompts are kept. | Retention becomes a billing question rather than a privacy one, and the default is somebody else's. | Decide retention deliberately, and prefer a design where your prompts stay on your infrastructure. |
Run this check tonight
- Ask your current tool for the single most expensive conversation last month.
- Ask it what one named customer cost you, against what they pay.
- Ask it what next month costs at your current growth, with a range rather than a point.
- Check what your plan's data retention actually is, and who chose it.
- Whatever goes unanswered is the gap, and the gap is the whole comparison.
Two products, four questions, read off both sites
Helicone's published plans and its acquisition post, read 2026-08-03, against what Culpa does. No pricing comparison is drawn, because the two are priced on different units and a like-for-like number would be invented rather than measured.
Both attribute cost, and the honest gap is narrower than a comparison page usually admits. Helicone routes, caches and is open source. Culpa adds margin against revenue and a forecast, and keeps the prompts on your own infrastructure rather than on a vendor's with a retention tier attached.
Every number, with its confidence and source
| Figure | What it means | Confidence | Source |
|---|---|---|---|
| $79 to $799 per month | Helicone's published Pro and Team plan prices, before usage-based charges | provider-reported | helicone.ai/pricing, read 2026-08-03. Both tiers state that usage-based pricing applies on top. |
What a generic answer can’t know
No comparison table can tell you which of these two answers the question you actually have, because that depends on your own traffic and your own unanswered questions. Run the free scan and see whether the conversation it names is one you could already have found. Culpa keeps your prompts and responses on your infrastructure, and counts the calls to run your plan.
Questions founders ask next
Is Helicone shutting down?
No, and it's worth quoting them rather than paraphrasing. Their post of 2026-03-03 says Helicone has been acquired by Mintlify and that services remain live for the foreseeable future in maintenance mode, with security updates, new models and bug and performance fixes still shipping.
How is Culpa different from Helicone?
Less than a comparison page usually claims, on attribution. Helicone publishes Sessions, User analytics and Custom properties on every tier, so it groups cost by conversation and user too. The differences worth choosing on are margin against revenue you enter, a forecast of next month from your own history, and prompts that stay on your infrastructure. Helicone's own strengths are routing, caching and being open source.
Do I have to send Culpa my prompts?
No. Culpa runs on your own infrastructure and your prompts and responses stay there. Culpa counts the calls to run your plan. That's the structural difference from any hosted observability layer, whose retention is set by the plan you're on.
Can I run both?
Yes, and some teams should. They answer different questions, and Culpa reconciles what it captures against your provider invoice regardless of what else sits in your stack.
On your infrastructure
Culpa runs on your infrastructure. Your prompts and responses never leave it. Culpa counts calls to run your plan, and it fails open, so if it ever breaks your app keeps running.
How Culpa works
Find the culprit. Not just the total.
Your dashboard shows what you spent. It stops short of who spent it. Culpa shows the conversation, the user and the feature behind it.
Your prompts stay local.
Culpa runs on your own infrastructure. What you send to a model reaches us at no point.
Every dollar has a name.
Follow any charge to the conversation, the user, the feature and the customer behind it.
See the bill before it lands.
Cost your next feature before you ship it. You get the likely bill and the worst case, at best, median, p90 and p99.
Three steps to your first answer.
Change one base URL.
Or drop in the Python or TypeScript library.
Find your most expensive conversation.
In the first session, not the first week.
Cost your next feature before you ship it.
Why the bill went up
Example dashboardCalls traced
418,209
across 3 projects
Spend this week
$378.41
+ $182 vs last week
Failed calls
312
74% retried, and you paid for all of them
+ $182 this week traced to one culprit
Spend over 14 days
Most expensive users
Next week forecast
Graded against reality. Accuracy shown as results land.
Free, no card, no account
Run the free Cost Leak Scan
It shows your most expensive conversation before you install anything.
Keep reading
Sources: Helicone: joining Mintlify, Helicone pricing. Last reviewed 2026-08-03. Plain text version.