Guides / price an unlimited ai plan
Pricing an unlimited AI plan against a tail you haven't measured
An unlimited AI plan prices a promise against a maximum nobody has computed, and on a long-tailed usage curve that maximum keeps moving. A generous included allowance with a stated overage reads nearly as well and bounds the risk. Culpa, a local-first LLM cost, margin, and forecast ledger, reports the spend distribution so the tail is a number rather than a fear.
Why this happens
Unlimited works in traditional SaaS because marginal cost is near zero. On an AI product marginal cost is the dominant cost, so unlimited transfers the entire variance of your usage curve onto your own margin. The damage is concentrated: a handful of accounts discover the ceiling doesn't exist and build workflows around that fact. Those accounts are usually your most engaged, which makes the conversation afterwards harder, not easier.
What this usually looks like
- Your unlimited tier has accounts costing multiples of what they pay.
- Nobody has computed what the heaviest plausible account would cost on the plan.
- The unlimited plan was priced against competitor positioning rather than your own unit costs.
- Usage on that tier grows faster than on metered tiers, which is the plan working as designed against 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 |
|---|---|---|
| Launching unlimited before measuring your own tail. | You've priced a promise against a maximum that doesn't exist yet and will keep rising. | Measure the 99th percentile first, price against that, and only then decide whether unlimited is affordable. |
| Matching a competitor's unlimited plan. | Their cost structure, model mix and usage curve aren't yours, so the same price means a different margin. | Price from your own measured distribution. A plan that works for them can be fatal for you. |
| Treating fair-use language as a cost control. | Fair-use clauses are rarely enforced and never enforced early, so the spend lands before the conversation. | Implement an actual ceiling with alerting, then handle exceptions commercially rather than in the terms. |
Run this check tonight
- Compute what your single heaviest account costs today, then double it. Can the plan absorb that?
- Find your 99th percentile account cost, not the 90th. Unlimited is priced against the extreme.
- Check whether usage growth on the unlimited tier outpaces your metered tiers.
- Ask what happens if your three heaviest accounts all double next quarter.
- Confirm you'd know within days rather than at month end.
What unlimited costs when the tail moves
Illustrative example
An unlimited tier at $200 a month on Claude Haiku 4.5 for volume work at $1.00 per million input and $5.00 output. Usage figures are modelled.
Growth stops rescuing an unlimited tier once the extreme share crosses a few percent, because each new typical account adds $160 of margin and each extreme one removes $650.
Every number, with its confidence and source
| Figure | What it means | Confidence | Source |
|---|---|---|---|
| $40 to $850 | modelled monthly cost range on a single $200 unlimited tier | estimated | Both endpoints at real Claude Haiku 4.5 rates per 1k tokens from the price book, effective 2026-07-02. A range because the usage profiles are modelled. |
| -325% to 80% | modelled gross margin spread from typical to extreme account on a $200 unlimited tier | estimated | ($200 - $850) / $200, from the teardown arithmetic at real Claude Haiku 4.5 rates. Modelled usage, so published as a scenario. |
What a generic answer can’t know
Nobody outside your own data can tell you the shape of your usage tail, and the tail is the entire risk in an unlimited plan. That needs per-customer spend over time. Culpa records it on your infrastructure, keeps your prompts and responses there, and counts the calls to run your plan.
Questions founders ask next
Is an unlimited AI plan ever a good idea?
It can work where your marginal cost per user is genuinely small and bounded, or where the plan is a deliberate loss leader you've sized. It fails when it's chosen to match a competitor without measuring your own tail first.
What's the safer alternative to unlimited?
A generous included allowance with a stated overage rate. It reads nearly as well in marketing, it bounds your risk, and it gives heavy users a clear signal before they become a problem rather than after.
Does fair-use language protect me?
In practice, rarely. Fair-use clauses are enforced late if at all, so the spend has already landed by the time anyone invokes them. An actual ceiling with alerting does the job the clause is imagined to do.
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: Anthropic pricing. Last reviewed 2026-08-01. Plain text version.