Rocketship today introduced a $12 per month Builder plan and changed how editing is billed. Small cosmetic changes, meaning colour, copy, type size and spacing, no longer consume a build credit. Every plan carries a monthly allowance of them, from 10 on the free tier to 400 on Mission Control. Users can also now choose how much reasoning goes into each build, and pay accordingly.

The $12 Builder plan

Builder is the app builder on its own: 30 build credits a month, 40 free cosmetic edits, up to 25 live apps, a custom domain, and Stripe Connect built in so an app can take money on day one. It does not include the AI workers. Sophie, Claire and Max stay on Launch and above, along with the voice receptionist.

It is a deliberately narrow plan. Most people arriving at an AI builder want to find out whether the thing in their head can exist. They do not need an outbound email agent on day one, and they should not be paying for one. Builder is that plan, at the lowest entry price in the category.

Why cosmetic edits stopped costing a credit

Watch anyone use an AI builder for an hour. The first prompt creates the thing. Almost every prompt after it is small: make that heading bigger, try a darker green, change "Get started" to "Book a call", add more space above the form.

Those are the moments where taste actually happens, and on a per-prompt pricing model they are also the moments where a meter starts running. People stop adjusting. They ship something slightly wrong because the third revision felt like it cost money.

Charging for a colour change is charging someone for having taste.

The reason we can do this is that we rebuilt how those edits run. A small cosmetic change no longer ships the whole project to a large model and asks for whole files back. It routes to a small, fast model that returns a targeted patch, which costs us a fraction of a cent. There was never a good reason to bill for that.

Structural work still costs a credit, because it genuinely costs us one. A new page, a new section, a new data model, or a sweeping change across every page in the project is real work, and it runs on the full model. So does anything the small model turns out not to be able to handle. If a request starts as a tweak and turns out to be bigger, Rocketship escalates it to the stronger model rather than handing back something half finished.

To be precise about what "free" means here, because a lot of pricing pages are not: each plan includes a set number of cosmetic edits per month, and they do not draw down build credits. Free is 10 a month, Builder is 40, Launch is 150, Frontier is 250, Mission Control is 400. Past the allowance, edits fall back to costing a build credit like anything else. It is an allowance, not an unmetered promise, and we would rather say so than have someone find out in month two.

Choose how hard the AI thinks

Not every build deserves the same effort. A one page site for a plumber is not the same problem as an app with sign in, a database and a subscription. So build depth is now a choice, priced from what each one actually costs to run:

Blitz, 3 credits. The default. Fast, and unchanged from what Rocketship shipped before today.

Deep, 6 credits. A frontier reasoning model on every page. Better structure, better copy, better judgement about what a page needs.

Max, 15 credits. Full reasoning and a much larger code budget per screen. This exists for real applications: authentication, data models, subscription gating, payment flows. Those are problems with a correct answer, and reasoning is what finds it.

Most builders either hide this decision or expose it as a raw model picker. Neither is right. Hiding it means everyone gets the same effort whether they need it or not. A model picker asks people to understand something they should not have to. Rocketship asks a question a person can actually answer: what are you building?

The choice is always the user's. Rocketship never quietly upgrades a build to a more expensive tier on your behalf.

The same choice when you edit

The first prompt is not where a project gets built. Everything real happens afterwards, in the edits, and that is where most tools quietly hand you their cheapest model. So the depth choice now applies there too.

Standard, 1 credit. Everyday changes. Small tweaks remain free.

Deep, 4 credits. A reasoning model, with roughly seven times as much of your project in view. For changes that touch several files at once.

Max, 8 credits. Full reasoning across the whole project. For a refactor, a data model change, or threading sign in or payment state through screens that already exist. These are problems with a correct answer rather than a plausible one, and that is what reasoning buys.

Other tools in this category do vary how hard their model thinks. The difference is that they decide, and you find out what it cost afterwards. Here you decide, the price is on the control before you press enter, and it does not move.

What has not changed

Blitz builds cost exactly what they cost yesterday. Nobody pays more for the same output. Launch, Frontier and Mission Control keep their AI workers, their voice minutes and their lead handling, and every paid plan gets a monthly allowance of free cosmetic edits scaled to the plan.

Apps built on Builder carry a small Rocketship badge. Removing it is part of the white label option on Mission Control, our agency tier. Every plan, Builder included, ships with Stripe Connect, so apps can charge their own customers from the day they go live and the money goes to the owner's own account.

Available now

The Builder plan is live today. Existing customers keep their current pricing and pick up the cosmetic edit change automatically.

See plans and pricing, or start from a finished template for your industry. Every template ships with lead capture already wired in, phone number included.

For press inquiries: support@deployrocketship.com