Checker · AI app builder

Credit ledger

What a build costs is read from the builder's own meter, before and after every prompt, into one ledger. Credits, tokens or dollars, whatever the builder counts in; where a builder shows no meter at all, the ledger keeps the prompts and their times.

When it runs: Around every prompt of the build
Used in: 21 reviews
Runs on: The builder's usage page, one line per prompt

Builders sell credits, and the review has to say what the brief's app cost in them. The ledger is one file per review with one line per prompt: what was asked, what the meter said before and after, how long the builder took, what it produced and whether anything already working broke. The pricing section's figures come from this file.

How it works

Before a prompt, the builder's usage page is read and the balance kept as the pending reading.
The prompt is typed into the builder's chat and sent, and the time noted.
When the builder goes idle, the usage page is read again and the line is written: the prompt, credits before and after and used, the seconds it took, the result, and anything that regressed.
A spending guard reads the balance before each prompt on pay-as-you-go builders and refuses to send under a floor, after one builder billed well past its plan.
The ledger feeds the cost figures in the pricing section and the build timeline.

What it records

Per prompt: Time, test id, the prompt, the meter before and after, the amount used, seconds taken, the result and any regressions.

What it cannot catch

The cost is as precise as the builder's meter: some show rounded figures, some lag a prompt behind, and some show nothing, in which case the review gives a prompt count.
Each builder's usage page is read by its own small reader.
Anything typed into the app by hand rather than prompted is noted apart from the ledger.

Where it was used

Every published review that ran this checker, with the figure it produced there, read from the review’s own data. Each figure opens the test results behind it.

Reviews that used the credit ledger
Review	Figure	Tested
Adalo	19 prompts typed	Invalid Date
Anything	23K credits used, to zero	Invalid Date
Atoms	96 / 115 credits used	Invalid Date
Base44	47.6 / 100 credits used	Invalid Date
Blink.new	100 credits used, of 105	Invalid Date
Bolt.new	5.5M / 10M tokens used	Invalid Date
Bubble	961 AI credits used, of 1,000	Invalid Date
Draftbit	2,823 / 3,000 free credits used	Invalid Date
Dreamflow	87% of the free credits used	Invalid Date
Emergent	50 / 50 free credits used	Invalid Date
Famous.ai	$96.72 billed, on a $7 plan	Invalid Date
FlutterFlow	2 AI requests I made	Invalid Date
Noloco	about 18 of 35 free credits used	27 Oct 2001
RapidNative	10 / 10 free credits used	Invalid Date
Replit	$5 / $20 credits used	Invalid Date
rocket.new	20 / 20 free credits used	Invalid Date
Rork	95 credits used of 130	Invalid Date
Softgen	75 credits used of 75	Invalid Date
Softr	~8.8 AI credits used	Invalid Date
Thunkable	2,000 tokens used of 2,000	Invalid Date
Zite	25.5 of 150 credits used	Invalid Date