Guides / culpa vs portkey
Culpa vs Portkey
Portkey is an AI gateway with observability, routing and granular budget limits that block spend at a ceiling you choose. Culpa, a local-first LLM cost, margin, and forecast ledger, prices every call from a versioned price book and forecasts next month from your own history. A cap and a forecast answer opposite questions, so most teams want both.
Why this happens
The distinction that decides this one is small and it matters. Portkey sells budget and rate limits, including budgeting on virtual keys, and that's a genuinely useful control: you pick a ceiling and requests stop at it. What a ceiling can't tell you is where you'll land. It's a number you chose, not a number derived from anything, so it says what you'll allow rather than what you'll spend. A forecast runs the other way, taking your own history and projecting the month with a range. The second thing worth knowing before you build a cost practice on Portkey is retention. Its free tier keeps logs for 3 days and metrics for 30, and Production at $49 a month keeps logs 30 days and metrics 90. Those are reasonable numbers for debugging and they're short for finance, because a quarter-over-quarter question needs more history than any of those tiers hold.
What this usually looks like
- You set a budget cap and still can't say what the month will actually cost.
- Your cap gets raised every quarter and nobody can say whether that's growth or waste.
- Log retention runs out before your finance review starts.
- Requests were blocked at a ceiling and nobody could say which customer hit it.
- You know what you're willing to spend and not what a given customer earns you.
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 |
|---|---|---|
| Treating a budget limit as a spend prediction. | A cap is a decision, not a measurement. Hitting it tells you about your cap, not about demand. | Keep the cap for safety and forecast separately from your own history. |
| Setting the cap from last month's total. | Last month is one sample. Without a range you'll either block real traffic or never trigger at all. | Forecast a range first, then set the ceiling above the top of it. |
| Assuming log retention is the same as metric retention. | Portkey publishes them separately, 3 days and 30 on the free tier, so the detail expires long before the chart does. | Check both numbers and decide which questions you need the detail for. |
Run this check tonight
- Ask what your current cap is based on, and who last changed it.
- Find your log retention and your metric retention, which may differ.
- Try to answer what next month costs, with a range rather than a point.
- Name the customer whose traffic drove the last cap increase.
What an overage line does to a bill you thought was fixed
Portkey publishes Production at $49 a month for 100k recorded logs, then $9 per additional 100k. Overage pricing is honest and it makes the bill a function of traffic rather than a fixed line. Take a product growing steadily and price the platform fee across the year. Log volume is modelled. The rates are published.
Nothing here is hidden, and that's the point. The published number is the floor, and which multiple of it you pay is decided by traffic you haven't had yet. That's a forecasting question about your observability bill, sitting underneath the forecasting question about your model bill.
Every number, with its confidence and source
| Figure | What it means | Confidence | Source |
|---|---|---|---|
| $49 per month | Portkey's published Production plan price, before overages | provider-reported | portkey.ai/pricing, read 2026-08-03. Production includes 100k recorded logs a month with $9 per additional 100k requests, 30 days log retention and 90 days metric retention. Free Forever covers 10k logs with 3 days log retention. Enterprise is custom. |
| $85 per month | the same Production plan at a modelled 500k logs a month, once overages are included | calculated | $49 base plus 4 additional blocks of 100k logs at $9 each = $49 + $36 = $85, using the overage rate published on portkey.ai/pricing and read 2026-08-03. The 500k monthly log volume is modelled, not measured. |
What a generic answer can’t know
Portkey and Culpa overlap less than the word cost suggests. Portkey sits on the request path and does things Culpa deliberately doesn't: routing, fallbacks, load balancing, caching, and blocking a request when a key passes its budget. Those are controls. Culpa reads what happened and prices it, holds the revenue each customer pays you so it can report margin rather than spend, and forecasts next month with a range. It also keeps that history in a database on your own infrastructure, so how far back you can look is a disk decision rather than a plan tier. If you already run Portkey, the gap Culpa fills is the one its own pricing page points at: you can cap spend and you still can't say what a customer is worth.
Questions founders ask next
Does Portkey forecast LLM spend?
Its published feature set doesn't claim to, as read on 2026-08-03. It sells budget and rate limits, which stop spend at a ceiling you set. That's a control rather than a projection. Nothing on its site describes estimating next month from your own history.
How long does Portkey keep my data?
On the free tier, 3 days for logs and 30 days for metrics. On Production at $49 a month, 30 days for logs and 90 days for metrics. Enterprise is custom. Those come from its pricing page read 2026-08-03, and the split between log and metric retention is worth noting because the detail expires first.
Should I replace Portkey with Culpa?
Probably not. Portkey sits on the request path doing routing, caching and enforcement, which Culpa doesn't do at all. Culpa reads the result and answers the finance question: what did it cost, what does that customer pay, and what does next month look like.
What does Portkey actually cost?
Free Forever at 10k recorded logs a month, Production at $49 a month for 100k logs with $9 per additional 100k, and Enterprise at custom pricing. Published on portkey.ai/pricing, read 2026-08-03. The overage line is the part worth modelling before you commit.
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: Portkey, Portkey pricing. Last reviewed 2026-08-03. Plain text version.