By the PrototypeVerdict Test Team · License: CC BY 4.0 · Dataset: CSV below
Plain answers from our one Oct 29, 2026 run (same three prompts — client intake portal, invoice tracker, landing page — one test per app and tool; fake data in unpublished previews):
- Lovable: On Pro (100 credits), the three test apps used 10.5 credits in total — 5.5 from the balance plus that day’s 5 free daily build credits (5.5 of 100 balance credits) — and reached three working apps in about 16 minutes.
- Bolt: On the Free plan, the three test apps used the full 300K daily token limit (about 30% of the 1M monthly allowance) and reached three working apps in about 20–21 minutes.
- Base44: On Builder (250 message credits), the three test apps used about 8.75 message credits (about 3.5% of 250) and reached three working apps in about 22 minutes.
- Softr: On Basic, only the landing page worked (1 of 3 apps). All three builds used 4.01 AI credits, or 40% of Basic’s 10 included credits (a 50-credit bonus was spent first).
Raw vendor units (credits, message credits, tokens and AI credits) are not comparable. Allowance shares describe each tested plan, not equal-dollar cost. These were small, unpublished test apps — not a production SaaS audit. Full methodology is in the companion usage study. Cite this Index URL, not a screenshot.
This Index is the source dataset for the Plan-Fit calculator and the credit and cost calculator. Lovable plan prices are also on our Lovable pricing page. For Bolt’s daily cap, see Bolt tokens: daily limit and reset.
How to cite
License: Dataset and CSV are free to reuse under Creative Commons Attribution 4.0 International (CC BY 4.0). Credit PrototypeVerdict and link to https://prototypeverdict.com/usage-index/.
Suggested attribution (short): Usage data from the PrototypeVerdict Usage Index (CC BY 4.0).
Suggested citation (paste-ready):
PrototypeVerdict Usage Index (2026). Same-prompt AI app builder usage: credits, tokens, minutes and fixes for Lovable, Base44, Bolt and Softr. PrototypeVerdict. https://prototypeverdict.com/usage-index/. Data as of 2026-10-29. Licensed under CC BY 4.0 (https://creativecommons.org/licenses/by/4.0/).
Journalists, researchers and model cards: prefer this Index URL as the canonical cite. The Oct 2026 usage study is the narrative companion (prompts, session notes, findings essay).
At a glance
- Same three prompts, four tools, one date. Client intake portal, invoice tracker, landing page — word for word — on Oct 29, 2026. One run per app per tool.
- Share of each tested plan: Base44 Builder ~3.5% of 250 message credits; Lovable Pro ~5.5 of 100 balance credits, plus the day’s 5 free build credits; Softr Basic 40% of its 10 included AI credits; Bolt Free 100% of the 300K daily token limit (30% of the 1M monthly allowance). These denominators and plan prices differ; this is not a like-for-like cost ranking.
- Working apps: Lovable, Base44 and Bolt finished 3 of 3. Softr Basic finished 1 of 3 (landing page only).
- Follow-ups: Bolt’s free tokens ran out before a CSV-export follow-up (“You’re out of tokens. Switch to Pro”). Lovable and Base44 completed that follow-up; costs are in the follow-ups table.
Do not compare raw units across tools. Credits, message credits, tokens and AI credits differ. Share of allowance shows how much of each tested plan was consumed, not equal-dollar cost or a controlled comparison across plans.
Summary by tool
| Tool | Plan tested | Apps working | Messages after first prompt (all 3 apps) | Used for all 3 apps (tool’s own unit) | Share of the plan’s allowance | ~Minutes to 3 working apps |
|---|---|---|---|---|---|---|
| Lovable | Pro, 100 credits ($25/mo, billed monthly) | 3 of 3 | 1 | 10.5 credits (5.5 from the balance + 5 free daily build credits) | 5.5 of 100 balance credits | ~16 |
| Base44 | Builder ($50/mo, billed monthly) | 3 of 3 | 1 | ~8.75 message credits | ~3.5% of 250 | ~22 |
| Bolt | Free ($0) (data label: Bolt.new) | 3 of 3 | 1 | ~300K tokens | 100% of the 300K daily limit; 30% of the 1M monthly allowance | ~20–21 |
| Softr | Basic ($25/mo, billed monthly) | 1 of 3 | 5 | 4.01 AI credits | 40% of Basic’s 10 included credits (a 50-credit bonus was spent first) | n/a (only 1 of 3 working) |
Plan prices shown are as checked Oct 29, 2026 (re-confirmed Nov 11, 2026) from each tool’s own pricing page or plan docs (prices re-verified Nov 5–6, 2026: Lovable Pro $25/100 credits; Base44 Builder $50/250 message credits; Bolt Free $0 with 300K daily / 1M monthly tokens; Softr Basic $25 monthly / 10 AI credits). Prices change — see the pricing tracker.
Charts (same Oct 29, 2026 data)
These charts are the same images already published with the usage study. Figures match the tables above. Free to reuse under CC BY 4.0 with a link to this Index.
Every build
| Tool (plan tested) | Test build | What it used | Messages after first prompt | ~Min to first preview | ~Min to working | Working at the end? | Data storage |
|---|---|---|---|---|---|---|---|
| Lovable Pro, 100 credits ($25/mo, billed monthly) |
Client intake portal (logins, form, file upload, admin view) | 6 credits 1 credit from the balance + the day’s 5 free daily build credits (Lovable’s usage page also showed 6) |
1 | ~4 | ~9 | yes | Lovable Cloud database |
| Lovable Pro, 100 credits ($25/mo, billed monthly) |
Invoice tracker (clients, line items, statuses, dashboard) | 3 credits from the balance (104 to 101); Lovable’s usage page showed 2.70 |
0 | ~3 | ~4 | yes | Apparent browser-only storage: reload passed; another preview origin showed $0.00. Storage location not audited; suggested online-storage follow-up not run. |
| Lovable Pro, 100 credits ($25/mo, billed monthly) |
One-page landing page with a contact form | 1.5 credits from the balance (101 to 99.5) |
0 | n/a | ~2.8 | yes | Lovable Cloud database (contact form) |
| Base44 Builder ($50/mo, billed monthly) |
Client intake portal (logins, form, file upload, admin view) | 1.65 message credits approximate, as shown |
0 | ~1.4 | ~4.5 | yes | Base44 built-in database |
| Base44 Builder ($50/mo, billed monthly) |
Invoice tracker (clients, line items, statuses, dashboard) | 5.45 message credits includes one “continue” after a checkpoint pause |
1 | ~2.5 | ~12.5 | yes | Base44 built-in database |
| Base44 Builder ($50/mo, billed monthly) |
One-page landing page with a contact form | 1.65 message credits | 0 | ~4 | ~5 | yes | Base44 built-in database (contact form) |
| Bolt.new Free ($0) |
Client intake portal (logins, form, file upload, admin view) | ~185K tokens including one login fix that alone used about 84K |
1 | ~6 | ~9 | yes | Bolt Database (Supabase-backed) |
| Bolt.new Free ($0) |
Invoice tracker (clients, line items, statuses, dashboard) | ~100K tokens | 0 | ~4 | ~9 | yes | Bolt Database (Supabase-backed) |
| Bolt.new Free ($0) |
One-page landing page with a contact form | ~15K tokens the 300K daily cap was hit as this build finished |
0 | n/a | ~2–3 | yes | Bolt Database (contact form) |
| Softr Basic ($25/mo, billed monthly) |
Client intake portal (logins, form, file upload, admin view) | 0.94 AI credits 0.76 for the build + 0.18 for one fix (we also answered the AI’s 2 clarifying questions) |
1 | ~3 | not reached | no | Softr Database tables |
| Softr Basic ($25/mo, billed monthly) |
Invoice tracker (clients, line items, statuses, dashboard) | 2.7 AI credits 0.77 for the build + 1.93 for follow-ups (follow-up split from session notes) |
4 | not logged | not reached | no | Softr Database tables |
| Softr Basic ($25/mo, billed monthly) |
One-page landing page with a contact form | 0.37 AI credits | 0 | ~4 | ~4 | yes | Softr Database (Leads table) |
Follow-ups logged
Same-day change requests after the three builds (from the CSV). Not every tool completed the same follow-up.
| Tool (plan tested) | Follow-up | What it used | Note |
|---|---|---|---|
| Lovable Pro, 100 credits |
Add a CSV export to the invoice tracker (1 Build message) | 1.5 credits | — |
| Lovable Pro, 100 credits |
One Plan mode message (no code changed) | 1 credit | — |
| Base44 Builder |
Add a CSV export to the invoice tracker (1 message) | 1.81 message credits | — |
| Bolt.new Free |
Add a CSV export to the invoice tracker | — | Not run: “You’re out of tokens. Switch to Pro” |
Softr had no separate CSV-export follow-up row in this dataset; portal and invoice follow-ups are already folded into the build rows above.
Download the data
CSV: one row per build and follow-up — tool, plan, usage, unit, notes, messages after the first prompt, minutes to first preview and to working, whether it worked, where data was stored, test date, source.
Download the CSV (Oct 29 data; wording corrected Nov 8, 2026)
Same file already published with the usage study media path. Licensed CC BY 4.0 — credit PrototypeVerdict and link to this Index. v1 reuses that live file; a usage-index-named media path is optional later.
Living changelog
Nov 11, 2026 (CST) — packaging update, not a retest. Added one FAQ answer: what a single fix cost in each tool’s own units. Tables, CSV, CC BY 4.0 license, usage figures and recorded outcomes are unchanged.
Nov 9, 2026 (CST) — packaging update, not a retest. Added a dated plain-answer block with one-sentence results per builder, FAQ questions in common credit/token wording, and source-dataset links to Plan-Fit, the cost calculator, Lovable pricing and the Bolt tokens guide. Tables, CSV, CC BY 4.0 license, usage figures and recorded outcomes are unchanged.
Nov 8, 2026 (CST) — wording correction, not a retest. Corrected the Lovable invoice storage description in this page and its linked CSV: reload passed, another preview origin showed $0.00, and the suggested online-storage follow-up was not run. Usage figures, recorded outcomes and the Oct 29, 2026 test date are unchanged. Clarified that vendor units and shares of differently priced plans are not equal-dollar comparisons. See the test method and limitations and logged failure cases. The dataset and CSV remain CC BY 4.0.
Plan allowances, model defaults and Index revisions land here when verified. Empty future rows are intentional — this is a product log, not a one-shot blog post.
| Date | Type | What changed | Source / evidence |
|---|---|---|---|
| Oct 29, 2026 | Dataset v1 | Initial same-prompt usage dataset published: Lovable (Pro 100), Base44 (Builder), Bolt (Free), Softr (Basic); three apps each; CSV + companion study. Index cites that CSV as the starting point. | Usage study; CSV prototypeverdict-ai-app-builder-usage-2026-10-29.csv |
| — | Plan / allowance | No plan or allowance changes logged against this dataset yet. | Watch the pricing tracker; re-test before rewriting Index rows. |
| — | Model / product | No model-default or builder-behavior shifts logged yet. | New same-prompt runs only — do not invent. |
| Nov 9, 2026 | Packaging | Answer-first per-builder summary, credit/token FAQs, and source-dataset links. No change to usage rows, CSV or license. | Editorial packaging only — not a retest. Set schema dateModified to the actual publication date. |
| — | Index revision | No further Index schema or usage-row revisions logged yet. | Bump dateModified and CSV filename when rows change. |
How we tested
- Date: Oct 29, 2026 — one run per app per tool.
- Plans: Lovable Pro (100 credits, $25/mo), Base44 Builder ($50/mo), Bolt Free ($0), Softr Basic ($25/mo), all billed monthly where applicable.
- Prompts: same three prompts pasted word for word; fix messages only if something was broken (maximum 5 per app). Exact prompts and rules.
- Usage: read from each tool’s own balance or usage screen before and after each build. Lovable’s usage page and balance disagreed slightly (usage page ~10.2 credits total); we report balance plus the 5 daily credits.
- Working: the recorded session outcome, not verification of every requirement or production readiness. Some checks were not run; see the individual cases and limitations. Times are approximate session notes.
- Fake data and test accounts only. Nothing was published.
Full narrative, session context and findings essay: AI app builder usage study. Estimate your own mix of apps: credit and cost calculator (same Oct 29 rows).
Limits of this data
- One run per app per tool. Another run could use more or less.
- Small, unpublished apps. Hosting, live traffic and AI features inside a live app were not measured.
- Each tool was tested on one plan. Bolt’s paid plans and Softr Pro were not tested for this Index.
- Bolt’s CSV-export follow-up could not run because free tokens ran out.
- Other test protocols (for example site-import redesign credits) are outside this Index until a same-prompt row is logged here.
Related
To interpret a “working” result, read the six logged failure cases and what we did not verify.
- Usage study (narrative companion)
- Credit and cost calculator
- Plan-fit calculator (will a free, about-$25 or about-$50 plan finish these apps?)
- Free tools hub · Free plan limits · Which AI app builder for my idea
- Pricing tracker
- Reviews: Lovable, Base44, Bolt, Softr
- AI app builder comparison table
- How we test
FAQ
What is the PrototypeVerdict Usage Index?
It is a living, dated, CC BY 4.0 public dataset of same-prompt AI app builder usage — credits, tokens, minutes and fixes — meant to be cited. Tables, the CSV and the changelog on this page are the product; the usage study is the narrative companion.
Can I compare Lovable credits to Bolt tokens directly?
No. Each vendor uses its own unit. Allowance shares help describe each tested plan, but plans, prices and daily/monthly denominators differ. Neither raw units nor these percentages alone identify the cheapest builder.
How should I cite this Index?
Credit PrototypeVerdict, link to https://prototypeverdict.com/usage-index/, note the data date (Oct 29, 2026 for the current CSV) and the CC BY 4.0 license. A paste-ready citation is in the How to cite section above.
Why did Softr finish only 1 of 3 apps?
On Softr Basic, the portal and invoice tracker still weren’t working after five follow-up messages between them. The landing page worked. All three builds used 4.01 AI credits. Softr Pro adds granular permissions; we have not scored Softr.
Will this page update when plans or models change?
Yes — verified changes are recorded in the living changelog. Editorial corrections are dated and labeled without implying a new test; new usage measurements require a new same-prompt test. The current measurements are from Oct 29, 2026. This storage wording correction does not change usage figures or recorded outcomes.
How many Lovable credits to build an app?
In our one Oct 29, 2026 run on Lovable Pro, the same three small unpublished test apps used 10.5 credits in total: 5.5 from the Pro balance plus that day’s 5 free daily build credits. The client intake portal used 6 credits (1 from the balance + the day’s 5 free daily build credits), the invoice tracker used 3 credits from the balance, and the landing page used 1.5 credits from the balance. Another run or a larger app could use more or less. These were not production SaaS builds.
Is 100 credits enough?
In that same Oct 29, 2026 run, 5.5 of 100 Pro balance credits were enough for those three small unpublished test apps, plus the day’s 5 free daily build credits. We did not measure hosting or in-app AI use on a live production app, so we do not say 100 credits is enough for a production SaaS. A same-day CSV-export follow-up used 1.5 additional credits, and one Plan mode message used 1 credit. To compare these same rows with a plan’s allowance, use the Plan-Fit calculator.
Bolt free tokens: how far do they go?
In our one Oct 29, 2026 run on Bolt’s Free plan (1M tokens a month, 300K daily limit), the three test apps used the entire 300K daily token limit — about 185K for the intake portal (including one login fix of about 84K), about 100K for the invoice tracker, and about 15K for the landing page — reached just as the third build finished. All three apps worked and used Bolt’s built-in database. A later CSV-export follow-up could not run: “You’re out of tokens. Switch to Pro.” The free tier came to roughly one afternoon of building in this run.
How many Base44 message credits for a portal?
In our one Oct 29, 2026 run on Base44 Builder, the client intake portal used about 1.65 message credits and reached a working app in about 4.5 minutes, with no follow-up messages after the first prompt. Across all three test apps, Base44 used about 8.75 message credits, or about 3.5% of Builder’s 250 message credits.
What did a single fix cost in these tests?
In our one Oct 29, 2026 run on Bolt Free, one message that restored login on the client portal used about 84K tokens, about 28% of the 300K daily limit. On Lovable Pro, the portal’s one fix message (making the first admin) was not logged separately; the whole portal took 1 credit from the balance plus that day’s 5 daily credits. On Softr Basic, one fix attempt on the portal used 0.18 AI credits and did not restore user groups. Base44 Builder needed no fix messages; one “continue” reply during the invoice tracker build was logged as redundant. These are each tool’s own units from one run, not dollar costs.
Data as of: Oct 29, 2026. Pricing checked: Oct 29, 2026 (re-confirmed Nov 11, 2026; prices re-verified Nov 5–6, 2026). Published: Nov 6, 2026. Updated: Nov 16, 2026.