Pricing

Pricing you can read.

One credit is $0.01 of AI usage at a published rate. You see an estimate and a cap before you send, a receipt after, and you never pay for MMNTM’s mistakes.

Free

Try the studio. Ship something small.

$0/ month

  • 300 credits to start
  • MMNTM Fast
  • 3 projects
  • Web + mobile previews
  • Export to GitHub

Pro

Recommended

For shipping every week.

$25/ month

  • 2,500 credits / month
  • MMNTM Pro + MMNTM Max
  • Unlimited projects
  • Publish to the web
  • Priority sandboxes

Team

Shared projects and pooled credits.

$30/ seat / month

  • 3,000 pooled credits per seat, per month
  • 2 to 5 seats
  • Shared projects
  • Roles & invites
  • Priority sandboxes

1 credit = $0.01 of AI usage at a published rate. Cost is computed from real usage and shown as a receipt on every message.

Top-ups: 1,000 for $12 · 5,000 for $50 · 20,000 for $180. Never charged automatically.

Questions

Seven things people ask.

Short answers. The long version is on screen in the Studio, on every message.

What is a credit?

One credit is $0.01 of AI usage. Cost is computed from real usage at a published rate (2.2×), and rounded up per turn. Every message shows a receipt with the tokens and the credits it settled at. Short follow-ups cost a few credits; a medium build usually lands in the tens.

Are fixes really free?

Yes. Heal rounds inside a run, failed-verification turns and platform failures cost 0 credits. Regression re-runs are free too, three per project per day, when the most recent change to the project is one MMNTM made in the last 24 hours and you have not edited the files since. You pay for work that reached you, not for MMNTM’s mistakes.

Which model does MMNTM use?

MMNTM has three tiers. MMNTM Fast handles quick edits and chat, MMNTM Pro is the default for building, and MMNTM Max brings the deepest reasoning for architecture-level work. Paid plans choose per message; every receipt shows which tier did the work, and MMNTM never routes you somewhere else silently.

Can I run it on my phone?

Yes. Every mobile project previews in a browser phone frame first, with safe areas and device presets. Scan a QR to open it in Expo Go on the same Wi-Fi, or build your own development client with expo-dev-client, which MMNTM scaffolds for you. The Run-on-phone dialog always shows which rung you are on and the next step.

Is there lock-in?

No. Projects are plain Vite or Expo repositories. Download a zip or push to GitHub from the Free plan. Bring your own Supabase, Neon or Convex with keys you hold. MMNTM hosts no proprietary backend, so there is nothing to migrate off.

Do unused credits roll over?

Plan credits roll over up to one month's allowance while you are subscribed, so a quiet month is not lost and a heavy one has headroom. Top-up credits never expire while you are subscribed. Twelve months after a subscription ends, whatever is left expires.

How do teams work?

Create a team under Settings → Team and invite up to five people. Team projects are shared, everyone builds in them, and every run spends from one pooled balance instead of anyone's personal credits. The Team plan is $30 per seat a month, and each seat adds 3,000 credits to the shared pool; until the team is on a plan, one-off top-ups fund the pool.

How do I cancel?

From Settings → Billing, through the Stripe customer portal. Pro-rated refunds within 14 days — see the Refund Policy linked in the footer. Your projects, code and history stay yours, and you can keep exporting them on Free (3 projects; the 300 starting credits are a one-off, so top up if you need more).

Comparison

Against Lovable and Base44.

Only claims from their own documentation, pricing pages and public reviews. Where we have nothing to say, we show a dash.

TopicMMNTMLovableBase44
Native mobile outputStandard Expo apps with genuine native modules, EAS-buildable on day one.Web only.Store packaging is, in its own docs, a lightweight native wrapper around a web view; no offline mode or HealthKit.
Verified before shownEvery turn passes typecheck and boot (plus a Metro bundle check for mobile) before the diff or preview appears.Fix-one-break-another loops; 10 free fixes per 24 hours; no refunds for platform errors.Whack-a-mole billing; completed 2 of 5 workflows in head-to-head tests.
Paying for the agent’s mistakesHeal rounds, failed-verification turns and regression re-runs cost 0 credits.Bills users to fix regressions the agent introduced.Bills users to fix regressions the agent introduced.
Price before you sendEstimate and hard cap on the composer, a token-level receipt on every message, at a published rate.Credit ranges are illustrations, not a cap; no cost shown before sending.No published top-up rate; no cost shown before sending.
Production when credits run outHosted output and OTA updates sit on a separate meter with a 14-day grace on failed payment.A unified balance pauses your live database, auth and storage at zero build credits.
Spend controlsA per-run ceiling, a daily ceiling that counts free rounds too, a loop detector that stops a run repeating itself, and an optional per-project monthly cap. The ceilings are on by default.Retry loops burned 16K–50K credits in minutes; compensation was 20 credits.
Lock-inPlain Vite and Expo repos; zip and GitHub push on Free; your own Supabase, Neon or Convex with keys you hold.Lovable Cloud exposes no service key or database URL and no migration path.Eject leaves an empty database; GitHub sync from $40, main branch only, and it can never reconnect.
Developer surface on the free planCode editing, per-version diffs, env vars, logs, versions and rollback.Dev Mode is read-only on Free.History is gated to Elite ($160+).
Model transparencyThree named tiers — MMNTM Fast, MMNTM Pro, MMNTM Max — with visible multipliers on paid plans; the run card shows which tier did what. No silent routing.Recent reviews blame quality drops on in-house routing.An in-house model exists for margin; the picker is gated to $40+.
Legible agent workStreaming plan checklist, diff cards, healing narration, gate strip and cost pill.A spinner, then a pile of changed files.A spinner, then a pile of changed files.
Secrets and privacySecret values never enter model context; a secret-in-bundle lint; no training on customer content on any tier.Training-on-by-default flip (September 2026); exposure of database credentials through a BOLA flaw.JWT leakage and PII in URLs.
Claims about Lovable and Base44 come from their public documentation, pricing pages and reviews as of September 2026. A dash means we make no claim.

Ready when you are.

300 credits to start on Free. No card.