Test log
Thunkable test data
Raw results for the Thunkable review. The verdict is in the review.
- Tested
- 27 October 2026
- Tester
- Miles Foster
- Account
- Bought for this review at the public price
- Tests
- 10
- Screenshots
- 12
- Method
- Published methodology
- Environment
- Google Chrome on desktop
Results
| ID | Test | Key measurement | Score |
|---|---|---|---|
| T01 | Prompt to Working App | Brief to a built app: About 5 min | n/a |
| T02 | Sign-in & Access Control | Sign-in screens: 0 | n/a |
| T03 | Database & Admin | – | n/a |
| T04 | Payments & Email Actions | – | n/a |
| T05 | Iteration & Credit Burn | Changes run: 4 of 12 | n/a |
| T06 | Publish, Speed & Mobile | A live app on Free: Not possible | n/a |
| T07 | Support & Cancellation | – | n/a |
| T08 | Spec Accuracy | Checks passed: 10 of 15 run | 3/5 |
| T09 | Code Export & Ownership | Code visible and editable: Yes | 2/5 |
| T10 | Pricing & Cost to Finish | Paid: $0 | 3/5 |
| Overall (in the review) | 2.7/5 | ||
Prompt to Working App
n/aNot Tested- Method
- The Fitting Room brief was pasted word for word (1,688 characters, still saying web app) into the prompt box on Thunkable AI's home screen at about 20:13 UTC on 27 October 2026, with the project set to Public, the free default.
- Brief to a built app
- About 5 min
- Requests
- 1
- Tokens used
- 400 of 2,000
- Screens made
- 13
- Remark
- The test ends at a live app with 25 of the 30 checks passing. No live app can exist on the free plan, so build speed is not graded. The time to a built app is reported as a fact.
Raw notes
The Fitting Room brief was pasted word for word (1,688 characters, still saying web app) into the prompt box on Thunkable AI's home screen at about 20:13 UTC on 27 October 2026, with the project set to Public, the free default.
Thunkable asked no questions. Its own timer shows 10 seconds to plan, 22 to sketch the main screen, 29 to design, 2 to read the request and 3 minutes 36 seconds to write the code, and the build was done about five minutes after I pressed Generate. It made 13 screens. The app first said it works on this device, and that sign-in accounts and shared data need a cloud database through my own Supabase account, and that until I connected one the app could be edited but would not run in the preview. I chose Keep local storage, and then the preview ran.
- The app ran only in the builder's preview, which needs a signed-in Thunkable account, so it has no public address.
Sign-in & Access Control
n/aNot Tested- Method
- I looked for sign-up, log-in and password reset screens, and for a way to tell a member, an instructor and an admin apart.
- Sign-in screens
- 0
- Attacks I could run
- 0 of 2
- Remark
- Sign-in needs a Supabase account that I did not create, so there is no grade.
Raw notes
I looked for sign-up, log-in and password reset screens, and for a way to tell a member, an instructor and an admin apart.
There are none. The Profile screen has two buttons, Instructor view and Admin view, and anyone who opens the app can press them: the admin screen with every enquiry, Manage classes (add, edit and delete) and All bookings opened with no sign-in. Thunkable says plainly that sign-in accounts need a cloud database, which means connecting my own Supabase account. Supabase's sign-up page carries a human check, and a new third-party account was outside this test, so I did not create one and could not run the two access attacks. Thunkable lists cloud database and authentication on the Free plan.
- The app I tested has open admin screens because it ran without a cloud database, which Thunkable's own message offers as the alternative.
Database & Admin
n/aNot Tested- Method
- I created records through the app, reloaded the preview and used the admin screens.
- Remark
- The six data checks need a database, so there is no grade.
Raw notes
I created records through the app, reloaded the preview and used the admin screens.
With local storage the app keeps its data on the device, not in a database: the Ask the studio message, the added class and the changed capacity all survived a reload of the preview. They also survived later changes by chat. Bookings could not be made, and the CSV export gave no file in the browser. A cloud database means the Supabase connection I did not make.
- Redeploy and export were not tried, and no admin data view of a real database was seen.
Payments & Email Actions
n/aNot Tested- Method
- Payments and emails need keys that are the owner's, so the six checks C24 to C29 are left out.
- Remark
- Not graded, as the methodology allows when the keys are not supplied.
Raw notes
Payments and emails need keys that are the owner's, so the six checks C24 to C29 are left out.
Thunkable said so itself: Stripe test payments and email delivery need credentialed integrations, and the app only records orders and enquiries. Pay $95 answered that checkout is not connected, no payment was taken and no credits were added. I created no Stripe account and entered no card.
- No payment or email was sent.
Iteration & Credit Burn
n/aNot Tested- Method
- Four of the twelve changes were run as chat requests: E03 colours, E04 About page, E05 capacity, E06 instructor filter.
- Changes run
- 4 of 12
- Worked first time
- 3
- Tokens per request
- 400
- Remark
- Fewer than 8 of the 12 changes ran, so there is no grade.
Raw notes
Four of the twelve changes were run as chat requests: E03 colours, E04 About page, E05 capacity, E06 instructor filter. I stopped when the tokens were at zero.
Each request cost 400 tokens, 20 percent of the free allowance, and took between 15 and 55 seconds. E03 turned the app dark green and cream. E04 added a screen named about, but I found no way to open it from the app. E05 set Power Reformer to 4 places, which the timetable shows. E06 added a filter by instructor that works: Tomás Reyes shows only his two Mat Flow classes. After the fifth request the tokens read 0 percent and an AI token limit reached page offered the paid plans. No change broke an earlier one.
- Chat edits to an app in the builder, not to a live app.
Publish, Speed & Mobile
n/aNot Tested- Method
- I opened the Publish menu, the publish checklist and the Download menu on the free plan.
- A live app on Free
- Not possible
- Screens in my app
- 14
- Free screen limit
- 5
- Test downloads left
- 2 of 2
- Remark
- There is no app to publish, so there is no grade.
Raw notes
I opened the Publish menu, the publish checklist and the Download menu on the free plan.
Publish iOS, Publish Android and Publish Web App all carry a crown. Publish Web App says to upgrade to the Builder or Advanced plan for responsive web apps, while the membership page lists responsive web apps under Advanced. The checklist read 5 of 23 ready and said the plan does not include publishing to the app stores. Publishing Android also stopped with a message that my project has too many screens for the plan: the app had 14 and Free allows 5. Accelerator allows 10 and Builder 100. The Download menu gives an Android test app that expires after 72 hours and an iPhone test app, 2 a month on Free. No public address exists, so no speed test, layout check or uptime monitor could run.
- I did not download a test app and did not try a phone. Nothing was uploaded to an app store.
Support & Cancellation
n/aNot Tested- Method
- I emailed three questions to support@thunkable.com on 27 October 2026: how to stop people signing up as admin, how to export data and code, and how to cancel a paid plan.
- Remark
- Not graded until a reply arrives.
Raw notes
I emailed three questions to support@thunkable.com on 27 October 2026: how to stop people signing up as admin, how to export data and code, and how to cancel a paid plan.
No reply had arrived when I wrote this. The membership page says to email billing@thunkable.com with questions and that you can change or cancel your plan any time. Account settings have a Delete account button and no other cancel control, which is what a free account shows. I paid nothing, so there was nothing to cancel.
- The cancel path of a paid plan was not tested.
Spec Accuracy
3/5Limited- Method
- The 30 checks were run by hand in the builder preview, not on a live app.
- Checks passed
- 10 of 15 run
- Checks not run
- 15
- Invented enquiries
- 3
- Seed facts on screen
- 10 of 10
- Remark
- Ten of the fifteen checks that ran is 67 percent, which is a 3 on the proportional scale. It is provisional, because nine checks could not run and none ran on a live app.
Raw notes
The 30 checks were run by hand in the builder preview, not on a live app. Six payment and email checks (C24 to C29) are left out because they need keys that are not mine. Nine more could not run, because booking needs a class credit and the only way to get one is the Stripe payment, which is not connected: C07 to C10, C12, C14, C15, C19 and C20 (the CSV button gave no file in the browser). Fifteen ran.
Ten passed: the timetable with the three classes, their times and instructors, places left on each class, the no-credit message (You need a class credit), the instructor's own classes, the admin adding a class that then shows on the member timetable, editing a capacity that then shows there, deleting a class, the count of bookings per class, and the Ask the studio form with the admin seeing the message. Five failed: sign-up, log-out and log-in, password reset, and the separate instructor and admin accounts (none exist), and the rule to invent nothing, because the admin screen shows three enquiries that I never wrote, from Maya Chen, Jordan Ellis and Riley Morgan, dated 25 to 27 October. All ten seed facts are on screen, the studio owner's name included.
- The grade describes the app in the builder preview with local storage, on the free plan.
Code Export & Ownership
2/5Limited- Method
- I opened the Code tab, the Download menu and the project menu on the Free plan.
- Code visible and editable
- Yes
- Source download found
- No
- Remark
- The code is visible but I found no way to take it out as files: 2. It is provisional, because I did not run it elsewhere and support has not answered.
Raw notes
I opened the Code tab, the Download menu and the project menu on the Free plan.
The Code tab shows the app's code, and it can be edited and saved on Free. I found no button to download it as source files. The Download menu makes an installable test app instead, and the Android file for sharing outside Google Play is not on the Free plan. I did not run the code anywhere else, did not read the terms and could not say who owns the app.
- The grade describes the free plan. The paid plans were not tested.
Pricing & Cost to Finish
3/5Limited- Method
- The pricing page, the membership page, the token meter and the plan limits were read on 27 October 2026.
- Paid
- $0
- Tokens used
- 2,000 of 2,000
- Remark
- The price is clear and nothing billed, but the free tokens ended long before the test did and the running cost is unknown: 3. It is provisional, because no paid plan was tried.
Raw notes
The pricing page, the membership page, the token meter and the plan limits were read on 27 October 2026.
Free is $0 with 2,000 AI tokens in all. The first request, the whole 13-screen build, cost 400, and so did each change, so the free tokens covered five requests. The same test needs about thirteen, around 5,200 tokens. The paid plans are $19 a month for 20,000 tokens and 1 live published app, $59 a month for 50,000 tokens, and $189 a month for 100,000 tokens. Accelerator allows only 10 screens a project, so my 14-screen app would need Builder. Tokens stop at zero and I saw no overage or auto top-up. The cost to run a live app is unknown.
- Yearly prices were shown on the membership page and not recorded here. A live app's running cost is not known.
Coverage
Tested
- One brief typed word for word, with no questions asked back
- The app in the builder preview: timetable, class page, pass page, profile, instructor view, admin view, class form and the Ask the studio form
- Four changes by chat (colours, an About page, a capacity change, an instructor filter)
- The Publish menu, the publish checklist, the Download menu and the Code tab
- The token meter, the membership page and the pricing page
- One support email with three questions
Not tested or limited
- Tested on the Free plan only: publishing, a public address, the app stores and the paid token allowances were not tried
- No live app exists, so there is no test app, no uptime record and no speed test
- The app has no sign-in. Sign-in needs a Supabase account that I did not create, so the logins, the two access attacks and the database were not run
- Payments and emails need keys that are yours, so they are left out
- Free tokens ran out after five requests, so eight of the twelve changes were not run
- Support is not graded: the email was sent on 27 October 2026 and no reply had arrived when I wrote this
- I did not read the terms and did not read what other users say
Still to test
- Publishing on a paid plan, a public address, speed and uptime
- Sign-in, the database and the two access attacks with a Supabase account
- The remaining eight changes
- The source export, run elsewhere
- A reply from support and cancelling a paid plan