PRICING

We have not priced this yet.
So this page prices nothing.

Memoth is in its design-partner phase. A tier table today would be a guess with a currency sign in front of it, and you would find out it was a guess at renewal. So this page gives the three things we do know: what a design partner gets, what pricing will meter once it exists, and what we measure for you either way.

Design partners

We're choosing a handful of teams to build with.

What you get

The memory layer under your agents early, tuned to your domain first. We measure token cost, task success and served-stale rate, and hand you the numbers. If they don't show up, you walk with a free audit.

Founder-direct iteration. Your corpus never leaves your environment.

How pricing will work

Three dimensions get metered. Headcount is not one of them.

When pricing lands it will meter the memory layer itself, priced per tenant and capacity tier, not per seat. Adding a teammate will not move the number. Keeping more of your corpus resident, and asking more of it, will.

FRAGMENTS STORED

What extraction wrote and governance kept, measured in fragment-months: one fragment held for one month is one unit. Fragments you delete stop metering from the month you delete them, and what governance rejected never metered at all.

The size of your corpus is not the meter. What survives governance is.

READS

Every promotion request an agent puts to your store. An abstention is a read: deciding that nothing cleared the floor is work Memoth did, and it lands on the same line as the answers rather than hiding in a free tier.

Counted per request, not per fragment returned.

WRITES

Every fragment written or rewritten — the ingestion runs that bring a source in, and the write-back of what agents learn while working.

A rewrite is a write. Re-reading the same fragment is not.

Three dimensions and no others. If a fourth ever shows up on an invoice, it shows up here first — that is the commitment this page makes in place of a number.

Measured, not promised

What a partner is owed before anyone is billed.

The partner terms above compress into four numbers. Unpacked, because “we measure it” is worth less than naming the measurement and saying what happens when it goes the wrong way.

Token cost
Tokens spent per task with the memory layer in front of your agents, against the same tasks without it. Measured on your traffic, not ours.
Task success
How often the agent finishes the job correctly, judged on tasks your team writes, because ours are not your work.
Served stale
How often an agent was handed a fragment that had already been superseded. Governance either keeps this near the floor or it is not doing its job.
Free audit
If those three do not move, you keep the audit — the corpus map, the failures we found and where they came from — and owe us nothing for it.

The same suites we publish results from, run against your corpus, handed to you as a table. Whether the table flatters us is not something we get to edit.