Pricing
A million rows a month, free.
Pay only past that.
One credit generates 10,000 rows. Every account gets 100 of them a month without a card. Buy more only when your work outgrows that, and if a run fails the credits go back automatically.
100 credits every month, which is 1M rows. No card, and no feature withheld.
- Every generator, every declaration, no feature held back
- Referential integrity checked and reported on every run
- Design on the canvas, from a sentence, or from your own schema
- CSV, JSON, SQL and Parquet export
- Bring your own AI key for schema design
A failed run costs nothing. Credits are refunded on the run itself, automatically, and you never have to ask.
Your AI key stays yours. Schema design uses your own provider key, sealed to your account and never readable by us. We do not resell somebody else's model.
Credits meter volume, not capability. Every guarantee works on the free tier.
Credit packs
One-off, no subscription, and they do not expire. Buy one when a project needs more than the monthly free credits.
A few real projects, or one big one.
Everyday work on staging databases and demos.
Large schemas and daily CI seeding.
Packs are bought from inside the app, where your balance lives.
Will the free tier cover you?
Set your real workload. One credit generates 10,000 rows, and every account gets 100 credits a month for free.
You would need 400 credits beyond the free 100. The Starter pack, $9, covers that.
Credits are only spent on rows that arrive. If a run fails, the credits go back automatically, and you can see every movement in your account.
Questions
- What is a credit?
- One credit generates 10,000 rows. A run costs one credit per 10,000 rows it produces, rounded up, so the smallest run costs one credit. Every account gets 100 credits a month for free, which is a million rows.
- What happens if a generation fails?
- The credits come back automatically, on the same run, without you asking. You paid for rows, and if the rows did not arrive you did not pay. Every movement is listed in your account so the balance can always be explained.
- Why credits instead of a monthly subscription?
- A subscription is a promise that the service is available all month. Misata is young enough that promising units of work delivered is more honest than promising uptime. When that changes, a subscription can too.
- Do credits expire?
- Purchased credits do not expire. The free monthly credits are added each month rather than resetting your balance, so an unused month is not lost and buying credits never wipes them.
- Do I pay for AI schema design?
- No. Schema design uses your own provider key, so you pay your provider directly at their price and we never mark up someone else's model. Credits are for generation, which is the part that costs us money to run. If demand grows enough to justify it, a hosted model tuned for schema design is the obvious next step, and it will be an option rather than a replacement: bring-your-own-key is not going away.
- Can I ask for something that is not here?
- Yes. If you need a dataset built to a specification, a volume the packs do not cover, or a guarantee the language does not express yet, say so. Early requests are how the roadmap gets set, and several declarations in the engine exist because somebody asked.
- Is anything held back from the free tier?
- No feature is. Every declaration, every guarantee and every export format works on the free tier. Credits meter volume, not capability, because a guarantee you can only afford at the top tier is not much of a guarantee.
- What does Misata not do well yet?
- It's listed rather than hidden. Locale-specific priors, free text that reads as templated on close reading, and a few other real edges are documented on the limitations page, found by testing our own output rather than assumed away. Read the limitations page.
Need something the packs do not cover?
A dataset built to your specification, a volume beyond the largest pack, or a guarantee the language cannot express yet. Tell me what you are trying to make true and I will tell you honestly whether Misata can do it, including when the answer is no.
On being early
Misata is built by one person, and it is early enough that the people who pay for it now are doing something more than buying rows. They are funding the hours that go into the engine, and they are the clearest signal I get about which guarantee to build next.
That is worth saying plainly, and it is worth not overselling. Buy credits because the tool does something you need, not to be kind. If it does not do that yet, tell me what is missing instead. That is more useful to both of us, and the free tier is there precisely so you can find out before spending anything.
The library is free, and always will be
Everything Studio does, the MIT-licensed engine does from the terminal with no account and no limit. Credits pay for us running it for you, not for permission to use it.
pip install misata misata generate --config schema.yaml --output-dir ./dataRead about the engine