Brain training apps that are not free: 2,000 reviews

We make a daily arithmetic app, so this comparison is not neutral and we are saying so before the first number. It is called Ponderanda, it is free, and it was built against the complaint this page counts. Read everything below knowing that.

There is a second thing to say first. We built Ponderanda before we had a record like this. We had read that brain-training apps cap free play at about three games a day, and we built against that without counting how many people minded. Ponderanda shipped on 17 August 2026. These reviews were pulled on 16 October. So this is the count we should have made first, made two months late, and it is as much a check on our own choices as on anyone else’s.

What we read

The 400 most recent Google Play reviews for each app, in English, from the US store, sorted newest rather than by Play’s default relevance so the window is reproducible.

App	Installs	Store rating	Window	Negative, with text	Developer replies
Lumosity	10,000,000+	4.79 of 324,580	29 Jul to 15 Oct 2026	29	63
Elevate	10,000,000+	4.57 of 519,860	7 Aug to 15 Oct 2026	93	115
Peak	10,000,000+	4.37 of 516,201	29 Sep 2025 to 14 Oct 2026	103	299
NeuroNation	10,000,000+	4.56 of 480,094	10 Jun to 15 Oct 2026	38	48
Mental Math: Everyday Speed	1,000,000+	4.38 of 8,124	May 2021 to Oct 2026	55	13

2,000 reviews, of which 318 are one or two stars carrying text. Only those 318 are counted below.

The four brain-training apps produce 400 reviews in two to four months, except Peak, which took a year. Mental Math: Everyday Speed took five years, so its complaints are older and describe older versions.

How the counting works

We read all 318 and chose the themes. A script then counts them with keyword rules, so the count can be re-run and argued with. A review can count under more than one theme. A keyword can misfire, so treat each count as close rather than exact, and where we read every match for a theme we say what we found.

Theme	Reviews	Share of 318	Mostly from
Pay to play, or free in name only	150	47.2%	Peak 61, Elevate 46
Ads in the free version	53	16.7%	Peak 30, Elevate 16
A free trial that charged, or a card before playing	41	12.9%	Elevate 21, NeuroNation 10
Accounts, sign-in, or a purchase that did not carry over	33	10.4%	Peak 12, NeuroNation 12
Bugs, crashes, will not open	17	5.3%	Peak 9
Text too small to read	12	3.8%	Mental Math 11
Timed questions	8	2.5%	Elevate 4, Mental Math 3
Doubts about the brain claims	3	0.9%	one each

Nearly half is about paying

150 of 318. We read every match. Two are not about money: one is about how little there is to play, and one is a two-star review that recommends the free tier. The other 148 are about price, a paywall, an app that says free and is not, or something the reviewer had paid for that then stopped working.

The cap we built against is there, and it is tighter than the three games a day we had read about. On Lumosity 10.20.24:

It used to be fun to play the free version as you had access to three games at a time that would change day to day. Now they’ve made it so you can only play a handful of lower levels on the free games, essentially meaning it is not possible to play with the free version of the app anymore.

On Peak, with no version recorded:

Don’t try to be slick now, Peak. From 4 daily workouts to 2 without watching ads.

The sharpest version is a test that asks for money at the end rather than the start. On Peak 4.31.9:

After taking a long test, I was asked for money to see the report. It’s the kind of trick I don’t like.

Charging for software is legitimate and we are not going to pretend otherwise. What the record shows is narrower: people object less to the price than to finding out about it late, after the download, after the sign-up, after the test.

That is also where the free trial theme comes from. 41 reviews say a trial charged them, or that the app wanted a card before they could play. On Elevate, with no version recorded:

You have to enter your email before they tell you it’s $15/month.

Ads are the price of the free tier

53 reviews are about advertising, and 30 of them are Peak’s. Most are not about ads existing. They are about ads that cannot be closed, or that decide when the game continues. On Peak, with no version recorded:

decent training but when an ad plays there is no way to exit the app until it is through. they hold you hostage to the add. Don’t lock me out of my phone. I Uninstalled your app.

One thing in Peak’s favour, because the reviews alone would hide it: its developer replied to 299 of these 400 reviews, 74.8%, far more than any other app here.

The clock, which matters more than its count

Only eight reviews are about timed questions. We read all eight, and seven are about the clock itself; the eighth is about tap targets in a speed test. Eight is small. We give it a section because it is the one complaint that names the person Ponderanda is for. On Elevate 5.258.0:

The stress from having a timer increased my anxiety that I knew Elevate wasn’t something I’d be interested in.

On Elevate 5.253.0:

This app would be better if it didn’t add timers on most things, forcing you to make quick decisions before actually thinking through things.

And on Mental Math: Everyday Speed 3.2.5:

there is a 10 second timer on every problem and the app limits how often you can pause the timer.

Almost nobody complains about the science

Three reviews of 318 doubt the brain claims. On Elevate 5.257.0, one reviewer says the games:

did not promote improved brain function as promised

That is worth knowing because it cuts against what we expected. The claims are the part of this category we were surest was wrong, and the reviewers barely mention them. The regulator did. In January 2016 the US Federal Trade Commission said Lumosity’s maker had deceived customers with claims that its games would improve performance at work and school and delay age-related cognitive decline. It agreed to pay $2 million, against a $50 million judgment suspended because of its finances. People who download these apps mostly do not argue about whether they work. They argue about what they cost.

What this says about Ponderanda, including where it fails

Ponderanda is five arithmetic problems a day, the same five for everyone, with one attempt. Against this record:

Paying. There is nothing to pay for. No subscription, no trial, no card, and every past day stays playable for free.
Accounts. There are none, so there is nothing to sign in to and nothing to lose when you change phones, apart from the history on the old one.
The clock. Ponderanda records how long each answer took, and never shows a clock while you are thinking.
The claims. It never says it makes you smarter. It says you get faster at the arithmetic you practise, which is what the evidence supports.

And where it does a version of the thing these reviewers dislike:

It limits play. One attempt a day is the design, not a price, but it is a limit, and a wrong answer on the second problem ends your day.
A retry costs an ad. After a failed day you can watch a rewarded ad for one more go. That is close to Peak’s “2 without watching ads”. The differences are that you choose it, it never appears inside the day’s five problems, and it never changes the result that was recorded. A second rewarded ad lends you a colour palette for a day, and the same palettes can be earned by playing.
There is an ad you did not choose. On Android, an App Open ad can show when you start the app from cold. It never shows before the day’s attempt is done, never in your first three sessions, and never within four hours of the last one. It is still an ad you did not ask for.
The text size question is open. Eleven reviews of Mental Math: Everyday Speed say the numbers are too small to read. One of Ponderanda’s tests draws a notice at enlarged system text; none draws the problem screen that way. That is the next check this record asks of us.

Three honest limits on everything above.

This is one locale, US and English, and every brain-training app here has reviews in many more languages than this.

The fifth app is not a brain-training app, and its five-year window makes its reviews older than the rest.

And the counts come from keyword rules. We read every match for the paywall, timer, claims and text-size themes, and not for the others. The files and the script that counted them are kept, so anyone disagreeing with a number can disagree with something specific.