Skip to content
AI Apps That Work

Deployment

Where to host your app, and what it costs to run: $0 to deploy and try it end to end on free plans.

Your app runs on Vercel or any host that runs Next.js, with a hosted Postgres database.

What it costs to run

$0 to build and try it end to end, on free plans:

ServiceFree planWhat the template uses
Vercel HobbyPersonal, non-commercial useEach server request can run up to 300 seconds; up to 100 cron jobs, each at most once a day. The template schedules three daily crons. Keep Fluid compute on (the default for new projects)
Neon100 hours of compute a month (Neon's CU-hours) and 1 GB of storage per project; sleeps after 5 minutes idleAn idle app lets it sleep, so trying it uses little of the free compute
Resend100 emails a daySign-in links. Until you verify a domain, it only delivers to your own address. Google sign-in sends no email
Google sign-inFree
StripeTest mode is freeTop-ups and subscriptions, with test cards
AI modelsA built-in placeholder model costs nothingTry every AI feature end to end before you add a key
File storageOptionalOnly needed once your app takes uploads

Real AI answers for $0. A Gemini API key from Google AI Studio works without billing, within the free tier's rate limits (shown in AI Studio). Google may use free-tier prompts and answers to improve its products, and people may review them: use it to try the template, not with your customers' data.

For storage, Cloudflare R2's free tier covers 10 GB, but it may ask for a payment method before you create a bucket. Any S3-compatible store works.

Once you charge customers, Vercel's free plan no longer applies: move to Vercel Pro (about $20 a month) or another Node host. Stripe live mode takes a fee per payment, with no monthly fee. Neon and Resend stay free until you outgrow their limits. Your AI spend stays within the monthly budgets you set.

On this page