What you can actually build on Shipd's free tier

Most "free" AI app builders are free in the way a demo is free — you get a few prompts, then a paywall. We wanted to write down, plainly, what Shipd's free plan actually gives you and where it stops being enough, instead of leaving that to a pricing page footnote.

What's actually on the free plan

Signing up doesn't ask for a card. You get:

• 100,000 tokens a month, reset monthly, no expiry pressure within the month
• All four frameworks — React, Vue, Svelte, or Vanilla, picked per project
• Live preview and one-click deploy to Vercel, so a project is a public URL the same session you build it
• Version history, so you can roll a project back if a prompt takes it somewhere you didn't want
• A real backend, not a mock — end-user sign-in and sign-up work the moment your generated app loads, and each user's data is isolated automatically. You don't wire this up; it's there because the app has routes that need it.

What's not included: GitHub push and custom domains are gated to the paid Standard tier, and things like QA test generation and full repo-aware generation (indexing an existing codebase and matching its conventions) sit behind the Codebase-Aware tier. Both of those tiers are listed on the pricing page as coming soon — right now, the free plan is the only one you can actually sign up for, which makes it worth understanding well rather than treating as a placeholder.

What 100k tokens actually covers

Token counts are an awkward unit to reason about until you've watched the meter move on a real project, so here's the honest shape of it.

A generation call spends tokens on two things: the context it has to read (your prompt, the existing project state, relevant files) and the output it writes (new or changed code). The first prompt on a new project — the one that scaffolds routes, layout, and initial components — is the expensive one, because there's no existing project to diff against. Every prompt after that is cheaper, because Shipd is generating a change against code that already exists rather than a whole app from nothing.

In practice that means:

• One real project, built and iterated on properly. A multi-page app — say a small SaaS dashboard with a landing page, a sign-in flow, and two or three authenticated views — plus a working round of follow-up prompts to fix layout, copy, and edge cases, fits comfortably inside 100k tokens for most builds.
• A handful of smaller projects, if you're prototyping — a single-page tool, a landing page, a form-and-results flow — each generated and lightly iterated, rather than one project taken through dozens of revision rounds.
• Where it runs out: the free tier isn't built for a project you keep re-generating from scratch, or for treating every small copy tweak as a full re-prompt instead of a targeted edit. Large multi-page apps with many rounds of "regenerate the whole thing" burn through the allotment fast, because each of those is closer to a first-prompt cost than a follow-up one.

The owner/admin view of a project shows the running total against the monthly limit as you build, so you're not guessing at it — you can see exactly how much of the 100k a given session used before you commit to another big prompt.

Getting more out of it

The cheapest way to stretch the free tier is to treat prompts like diffs, not rewrites. Ask for the specific change you want — "add a settings page," "fix the spacing on this card" — rather than re-describing the whole app and hoping it regenerates cleanly. Incremental prompts are what the pricing is actually built around; full re-generations are the expensive path regardless of which plan you're on.

It's also worth deciding early whether a project is a real build or a throwaway experiment. Free tier tokens reset monthly, but they don't roll over, so a month spent on three abandoned prototypes is a month you didn't spend finishing one.

If you outgrow 100k tokens regularly, that's a useful signal on its own — it usually means you're shipping something real, not evaluating a tool. The Standard and Codebase-Aware tiers exist for exactly that point, with 2M and 8M tokens a month respectively, though both are still marked coming soon on the pricing page as of this writing.

The honest takeaway

100k tokens a month, with no card and no time limit, is enough to build and deploy one real multi-page app with a working backend, or try a few smaller ideas — as long as you're prompting for changes rather than rewrites. It's not enough to treat as an unlimited scratchpad for a project you keep starting over. That's a real free tier, not a trial with a friendlier name, and it's worth using it like one: pick a project, build it properly, and watch the counter instead of guessing at it.