Pricing

There is one plan. The fee is a share of revenue you actually collected, which means Lath costs nothing in a month where you earn nothing.

Collected       across 412 payments            $10,000.00
Lath fee        1%                                $100.00
Requests        3,000,000 requests                  $1.35
CPU time        24,000,000 CPU-ms                   $0.72
DB compute      365 CU-hours                       $58.04
DB storage      12 GB-months                        $6.30
Email           27,000 messages                     $6.48
                                              ───────────
Invoiced                                      $172.89 USD

You keep $9,827.11 USD — 98.3% of what you collected.

Every figure on this page is in United States dollars (USD). USD is the currency Lath invoices in, the currency the fee ladder below is denominated in, and the currency payouts settle in.

The rate falls as you grow

One percent to $100,000 USD collected in a month, then half a percent above it. The rate falls as you grow rather than as you negotiate, and the tiering is applied to the account rather than to any single project.

First $100,000.001%                              $1,000.00
Next $150,000.000.5%                              $750.00
                                              ───────────
Fee on $250,000.00                            $1,750.00 USD   —  0.7% blended

Infrastructure is the list price, times 1.5

Every infrastructure rate on this page is the published list price for that unit, multiplied by 1.5, in USD. Not “about” 1.5 and not “a modest markup” — that exact number, on every line, with no line priced differently from any other.

Each unit is one a provider sells by name: a request, a CPU-millisecond, a CU-hour, a GB-month, a message. That is what makes the claim checkable rather than merely stated — you can look up the list price for any of them, multiply by 1.5, and get the number in the table. We do not tell you whose list, and that does not weaken the check: the benchmark is public and the comparison costs you one search.

We publish the multiple so you can check it. We do not publish what we pay, because that number would change without your price changing.

What this does not say is that 1.5× is our margin. It is the margin on the meter, and the meter is not where Lath makes its money — on the example above, the fee is $100.00 and everything the infrastructure markup adds is a fraction of it. The multiple exists so that the number you are billed for a database is traceable to a number somebody else publishes, not because it is the business.

Two costs are deliberately absent from the table. Egress is not billed at all, because it is not billed to us. And the fixed monthly cost of the platform underneath is not metered to anyone, because it does not vary with what you run — charging it per project would bill the second project for the first one’s existence. Both sit inside the fee.

One caution about the second figure. It describes a customer collecting twenty-five times the first one’s revenue while running five times the infrastructure, and that ratio is an assumption rather than a measurement — Lath has no customer at that scale to read it off yet. Usage grows more slowly than revenue for every hosted product we know of, so the direction is sound; how far is a judgement. The first real customer at that scale replaces it.

With no revenue collected

The fee is one percent of nothing. Infrastructure you are running is still billed, because it is still running — a dormant project with a database costs what the database costs.

Collected       no payments yet                     $0.00
Lath fee        1% of nothing                       $0.00
DB storage      5 GB-months                         $2.63
                                              ───────────
Invoiced                                        $2.63 USD

Limits refuse growth, never operation

A spending limit blocks new billable resources — a new service, a new database, a deploy that increases spend. It never stops serving what is already running, and neither does a failed payment. Production keeps serving while we retry, contact you, and give you dates in advance.