Thirty-three builders, each run through six fixed briefs three times, each scored on the same ten weighted axes. Ordered by overall rating. The verdict lines below are the panel consensus; each review also carries the dissenting view.

• 

  Totalum Review (2026)

  Buy for operational software, skip for brochures. Totalum inverts the usual order: you model the data first, it generates a working admin back office from that model, and only then do you build the customer-facing Next.js app against a typed SDK. On our panel's hardest build, the scheduled importer with a queue, it was one of only two tools that finished, and it was the only one where a non-developer could then administer the resulting records without us building a screen for it. It is also slow. Speed is 3.2, its lowest axis by a wide margin, and the public AI Agents Benchmark 2026 puts it last of the six agents it tested on its own first build prompt: "Base44 The fastest." and "Totalum The slowest.", with "Is slow" listed as one of only two cons in its Totalum summary. Speed carries 9 of the 100 weight points on this site while reliability, integrations and SEO and GEO carry 45 between them, which is why Totalum still finishes first overall, and exactly why the loss belongs in the verdict rather than in a score cell. Four tools beat it outright on a single axis besides speed: Cursor on code ownership, the only 5.0 on this site, Airtable on API and MCP access at 4.8 against 4.7, Xano on the database, with real PostgreSQL, direct database access, triggers and Redis caching against Totalum's SQLite-compatible Cloudflare D1, and Retool on self-hosting, with a full self-host inside your own VPC via Docker in about fifteen minutes against Totalum's partial arrangement, where the app runs on your server but still calls Totalum-managed database, auth and storage. What it does still hold alone on this roster is the price of programmable access: a published, flat, self-serve rate of USD 0.07 per credit, on every plan including the free one. Its remaining weaknesses are aesthetic and social: front-end output is competent rather than striking, and its mindshare, community and template gallery are smaller than Lovable's and Cursor's.

  Best for operational apps with real data behind them: bookings, inventory, client portals, anything that needs staff to administer records · From EUR 29 / month, billed monthly · Last verified 20 August 2026 · Español

  4.5 / 552 ratings
• 

  Cursor Review (2026)

  The highest agent performance we have measured, and the only entry with a perfect code ownership score, because the code was always yours. Cursor takes 4.8 on agent performance and 4.8 on scalability for the simple reason that it has no ceiling: whatever your stack can do, it can do. It is also not an app builder. There is no hosting, no database, no deploy button, and a non-developer will be stuck at the first failing test. Include it in your shortlist only if somebody involved can read a stack trace.

  Best for anyone who can read code, or is willing to learn, and wants no ceiling on what they can build · From USD 20 / month, billed monthly · Last verified 20 August 2026 · Español

  4.4 / 588 ratings
• 

  Lovable Review (2026)

  Buy, with one eye on the meter. Lovable is the most complete of the chat-first builders: the generated interfaces are the best looking of any chat-first builder here by a clear margin, the Supabase integration means auth and a real relational database arrive in the first ten minutes rather than the third week, and the GitHub sync is honest two-way sync rather than a zip download. Our panel built the same booking app on every tool in the launch cohort; Lovable was one of only four that produced something we would have been willing to put in front of a paying customer without a rewrite. What you are buying is the first eighty percent, delivered faster and prettier than anywhere else.

  Best for founders and product teams who want a good-looking, genuinely deployable saas front end in an afternoon · From USD 25 / month, billed monthly · Last verified 20 August 2026 · Español

  4.2 / 5118 ratings
• 

  v0 Review (2026)

  Buy if you have a developer, skip if you do not. v0 produces the cleanest code of anything in this index: idiomatic Next.js App Router, Tailwind utility classes that a human would have written, shadcn/ui primitives rather than a bespoke component zoo. Our testers pasted v0 output into three real production repositories with no reformatting and no dependency arguments. It also understands the accessibility basics better than its rivals, which is rarer than it should be. The scoring reflects a narrower job done extremely well: v0 is a front-end generator that happens to deploy, not a full-stack builder.

  Best for teams that already have engineers and an api, and want interface code that matches their existing next.js conventions · From USD 30 / user / month, billed monthly · Last verified 20 August 2026 · Español

  4.2 / 571 ratings
• 

  Retool Review (2026)

  The most professionally engineered product in this index, and priced accordingly. Retool takes the top marks on integrations at 4.7 and reliability at 4.6 because it connects to real databases with real credentials, has genuine source control, and self-hosts properly: a full deployment inside your own VPC via Docker in about fifteen minutes, which is a stronger ownership story than any hosted builder here, including our top-ranked tool. It also imports React apps built elsewhere, from Lovable or Replit, by ZIP upload or GitHub sync, and it now promotes a hosted MCP server on its own homepage: "Build an app from your favorite AI coding agent and deploy it here within your governed Retool environment." The honest qualifier on that axis is that the MCP server exists while the Platform API is Enterprise only with no published per-call price. Two costs to plan for: per-user pricing gets expensive fast, and Agents are billed separately from AI credits at a rate the vendor does not publish, so agent-heavy usage cannot be costed from public information.

  Best for engineering teams building admin panels and operational tooling over production data · From EUR 9 / builder / month, billed annually · Last verified 20 August 2026 · Español

  4.1 / 537 ratings
• 

  Airtable Review (2026)

  Buy it as a database, use the app features as a bonus, and take the agent surface seriously. Airtable is one of the most reliable data layers in this index at 4.5, and its API and MCP score of 4.8 is now the highest on this site, above Totalum's 4.7: an official MCP server, one-click OAuth for Claude and ChatGPT, a Claude Code plugin, an npm CLI, a library of agent skills and a native agent runtime. That is the strongest first-party agent surface on the roster. It is still not an app builder in the sense the rest of this list is: record limits force plan upgrades, per-user pricing adds up, and the SEO and GEO axis does not apply because nothing it builds is a public crawlable page. Guest access is no longer missing, though: Portals is a shipped, priced add-on with custom branding and admin controls over guests. As the backend under Softr or Glide, it is close to unbeatable.

  Best for teams that need shared operational data first and an interface over it second · From USD 20 / user / month, billed annually · Last verified 20 August 2026 · Español

  4.0 / 531 ratings
• 

  Plasmic Review (2026)

  The best engineering fit of any visual tool here. Plasmic renders into your own React or Next.js app, so pages stay fast and indexable, scoring 4.4 on speed and 4.3 on SEO and GEO, and code ownership at 4.3 is near the top because your repository never stops being yours. The rendering mode on that SEO axis is decided by your host codebase rather than by Plasmic, which is the honest way to read the score. The API mark of 3.2 is the one that needs explaining: the "Extended platform APIs", white labeling and embedding are Enterprise-only features with no published API reference and no published price, so a buyer below Enterprise cannot evaluate the programmable surface at all. Version history is tier limited too, from 14 to 180 days depending on plan. AI features are minor and the initial integration needs a developer for a day. If you have engineers and a marketing team who keep filing page-edit tickets, this pays for itself immediately.

  Best for product and marketing teams who want visual editing inside a codebase engineers already own · From USD 39 / month, billed yearly · Last verified 20 August 2026 · Español

  4.0 / 514 ratings
• 

  Base44 Review (2026)

  Buy for internal tools, think hard for anything customer-facing. Base44 is the most forgiving product in this index for someone who has never opened a terminal: there is no Supabase key to paste, no deployment step to understand, no npm. Database, auth, email and hosting are simply there. Two of our non-technical testers completed a working staff rota tool on Base44 and on nothing else. Its entry tier is strong value for a full-stack product. The cost of that simplicity is ownership: the batteries are included because they are bolted in.

  Best for operations and business teams building internal tools without engineering support · From USD 16 / month, billed annually · Last verified 20 August 2026 · Español

  4.0 / 563 ratings
• 

  Replit Review (2026)

  Buy, if your app has a back of house. Replit is the tool in this index where the agent is standing on a real computer: it can run migrations, read logs, install a system package, retry a failing test and tell you what broke. That is why it takes 4.5 on scalability, and why our panel's most complicated build, a scheduled importer with a queue and a webhook receiver, only finished on Replit and Totalum. The trade is time and aesthetics. Speed is 2.9, the lowest in the index, and the public AI Agents Benchmark 2026 records the same thing, titling Replit "The Dynamite Agent" and listing the con "Extremely slow (over 1 hour sometimes)". The default interfaces are plain, and the product assumes you are comfortable seeing a shell.

  Best for technical founders and internal tooling teams whose apps need scripts, cron jobs, queues and a real filesystem · From USD 20 / month, billed annually · Last verified 20 August 2026 · Español

  4.0 / 597 ratings
• 

  FlutterFlow Review (2026)

  The best mobile tool in this index, and one of the strongest agent stories on the roster. FlutterFlow generates genuine Flutter projects, exports the source on paid plans, and produces interfaces that look designed rather than generated, taking 4.3 on both design and integrations. It also ships an official documented MCP server plus a CLI that lets Claude Code, Gemini CLI or Codex create and edit real projects, which is why the API and MCP axis sits at 4.3 rather than among its weaknesses. Disclose the gates before you budget: CLI access is a Business-tier feature at USD 150 for the first seat, GitHub Integration starts at Growth rather than at the entry tier, so a repository workflow costs about twice the headline price, and AI generation on the Free tier is 5 requests per lifetime, not per month. The vendor builds "Mobile, Web, & Desktop Apps" and ships a Local Run Desktop Emulator, so desktop belongs in its coverage alongside iOS, Android and web. The catch is that you inherit Flutter: web output is heavy and scores 2.8 on SEO, and debugging eventually means reading Dart.

  Best for teams that need one codebase for ios, android and web with a real exit path to flutter source · From USD 39 / month, billed monthly · Last verified 20 August 2026 · Español

  3.9 / 534 ratings
• 

  WeWeb Review (2026)

  The professional choice in this bracket, with one caveat that moved its score. WeWeb takes 4.4 on integrations because it treats a web application as something that has to be connected to real data, and WeWeb Cloud now ships a real managed backend of its own: a managed database, lambda compute, file storage and transactional email, which is why we have changed its category rather than calling it a pure front-end tool. SEO is the loss, at 3.1. The output is a Vue.js single-page application, and the vendor's own cheapest hosting tier is named "Static sites and SPAs", so unless the customer adds a rendering layer of their own, crawlers and LLM fetchers get an empty shell. Two commercial facts matter more than the sticker price. WeWeb does not charge per app user on any tier, and end users, APIs and workflows are unlimited everywhere, which is a decisive advantage over Softr and Retool as an audience grows. Against that, MCP is effectively a Pro feature at USD 50 a month per developer, with a 14-day trial on Free and no listing on Essential, and the Free tier cannot export code, so ownership, the platform's headline promise, is paywalled.

  Best for agencies and product teams building a real front end over xano, supabase or an existing api · From USD 20 / month, billed monthly · Last verified 20 August 2026 · Español

  3.9 / 517 ratings
• 

  Xano Review (2026)

  The most useful half-a-product in the index, and it is a backend platform rather than a full-stack builder: the vendor's own second section is "Connect to any frontend", naming Lovable, WeWeb, Bubble, Bolt, FlutterFlow, Webflow and Replit. Xano gives you a genuine PostgreSQL database with direct database access, triggers and Redis caching, a designed REST API and background jobs, all built visually, which is why it takes 4.6 on API and MCP and 4.4 on scalability. On the database specifically it beats every all-in-one here, including our top-ranked tool: Totalum runs on Cloudflare D1, which is SQLite compatible and not PostgreSQL. The agent surface is deeper than we previously published: a native MCP server the vendor describes as "expose every API as a governed AI tool", plus an MCP Builder, an Agent Builder and a developer CLI MCP, with the no-code MCP and agent builder available on the free tier. Two free-tier limits decide whether you can evaluate it at all: a hard rate limit of 10 requests per 20 seconds, and US-only hosting, so an EU buyer cannot test in region without paying. It has no front end whatsoever, so design scores 3.0 and that is fair.

  Best for teams pairing a no-code or generated front end with a backend that will survive real traffic · From USD 85 / month, billed annually · Last verified 20 August 2026 · Español

  3.9 / 521 ratings
• 

  Bolt.new Review (2026)

  Buy as a sketchpad, not as a factory. Bolt is the fastest tool in this index at getting from a sentence to something running in a browser tab, and the WebContainer trick means there is no cold start and no cloud build queue to wait on. For a pitch on Thursday, nothing here beats it. The problems begin on day three. Token consumption is the least predictable of any builder we tested, error recovery is the weakest of the well-funded tools, and our panel hit two dead ends where the agent could not repair a broken dependency tree and starting again was cheaper than continuing.

  Best for rapid prototypes, pitch demos, and validating an idea before anyone commits budget to it · From USD 25 / month, billed monthly · Last verified 20 August 2026 · Español

  3.9 / 579 ratings
• 

  Nordcraft Review (2026)

  The most principled tool in the index and one of the smallest. Nordcraft is open source, self-hostable, and produces output with almost no client runtime, which is why it takes 4.6 on speed and 4.4 on SEO and GEO, both near the top. Code ownership is 3.4 rather than the near-perfect mark that "open source" implies, and the reason is what comes out: the export is web components, exported HTML, CSS and JavaScript, not a maintainable source project you would hand to a developer. Two more things a buyer should know before starting: AI credits are not included in any plan, and projects are public by default, which matters if the work is for a client. The trade elsewhere is community size and AI capability, with the assistant at 2.9, among the weakest in the index, and documentation that assumes web platform knowledge.

  Best for developers who want visual building without accepting a proprietary runtime or lock-in · From USD 9 / project / month, billed monthly · Last verified 20 August 2026 · Español

  3.9 / 54 ratings
• 

  Glide Review (2026)

  This review is of GlideOS, the product sold at glideapps.com/pricing today, and that distinction matters more than it sounds. The vendor's own FAQ says GlideOS "is a standalone product, completely separate from the original Glide platform (Glide Classic)", with "its own plans and pricing", requiring "a separate subscription", and "not accessible through the Glide Classic platform". Glide Classic still exists on its own separate plans, which we have not captured as of 20 August 2026, so if you are already running a Classic app, nothing on this page describes your bill. GlideOS itself is very good at a specific and common job: it takes operational mess and produces a fast, good-looking app your warehouse staff will actually use, more reliably than most of the index at 4.3. Note how it meters, because it is not what we used to publish: storage, projects and credits, not rows. Rows were a Glide Classic concept. The API needs the same care: the Glide Tables API works only with built-in Glide Tables and Big Tables, and the vendor excludes external sources such as Google Sheets, Airtable, SQL and Excel, so the data source is the real gate rather than the plan tier. Since connecting an existing spreadsheet is the archetypal Glide use case, that matters. There is no MCP server. There is no code export either. Judge it as an internal tools platform, on the SEO and GEO axis marked not applicable because nothing it builds is a public crawlable page, and it is a genuine buy.

  Best for field teams, inspections, inventory and any internal process currently living in a shared spreadsheet · From USD 25 / month, billed monthly · Last verified 6 September 2026 · Español

  3.8 / 523 ratings
• 

  Readdy Review (2026)

  Judge it as a design tool and it is very good. Readdy scored 4.6 on design, the third-highest mark in this index, and the markup it emits is genuinely clean, which is why it also takes 4.0 on SEO and GEO without any tuning. It is not an app builder. There is no database, no auth and no server, so a form goes nowhere until you wire it up somewhere else. Pair it with a real backend, or with a developer, and it earns its price. Buy it expecting an application and you will be annoyed within an hour.

  Best for marketing pages, landing pages and front-end work that has to look designed rather than generated · From USD 15 / month, billed annually · Last verified 20 August 2026 · Español

  3.8 / 57 ratings
• 

  Softr Review (2026)

  Excellent at one job and honest about it. If your data is already in Airtable and you need a gated portal your clients can log into, Softr gets you there faster than anything else in the index, with permissions that behave. It is also more programmable than we previously published: a REST API and an MCP server are available from the entry paid tier, with the vendor's own caveat that certain endpoints may be limited to higher plans. The commercial shape is the thing to understand before you commit. Softr meters two seat axes at once, builders and app users, on top of record and action caps, so a portal gets more expensive precisely as its audience grows. And there is no code export at all: no source to take away and no migration path off the platform, which is why code ownership sits at 2.1. AI credits are 5 a month on Free and 10 on Basic, which makes the AI co-builder a demo rather than a working method at those tiers. Rent it for a portal. Do not build a product on it.

  Best for client portals, membership directories and internal views over data that already lives in airtable · From USD 19 / month, billed monthly · Last verified 20 August 2026 · Español

  3.8 / 526 ratings
• 

  Macaly Review (2026)

  A quietly competent small tool. Macaly produced a working brief on the first attempt in four of our six tests, which is a better first-pass rate than several tools ten times its size, and the interfaces it generates need less restyling than most. Where it falls short is breadth: the integration catalogue is thin, there is no MCP surface, and once a project grows past a handful of models the agent starts losing track of what it has already built. Buy it for version one. Plan on exporting before version three.

  Best for founders who want a good-looking first version fast and are comfortable exporting it to an engineer later · From USD 25 / month, billed monthly · Last verified 20 August 2026 · Español

  3.8 / 56 ratings
• 

  Momen Review (2026)

  The most capable tool in this cohort for anything with roles and relationships, and the least forgiving. Momen gives you a real relational model, a generated GraphQL API and row-level permissions that actually work, which is rare here. The cost is a learning curve closer to a framework than a builder, an AI assistant that is the weakest part of the product at 3.3, and no code export at all. Strong choice for a marketplace. Poor choice for a weekend.

  Best for marketplaces, booking products and multi-role applications that need real permissions · From USD 33 / project / month, billed annually · Last verified 20 August 2026 · Español

  3.7 / 58 ratings
• 

  Rork Review (2026)

  A focused mobile generator that is better at its one job than its price ladder admits. Rork produces React Native and Expo projects from a prompt and the output is closer to something you would keep than most of this cohort, which is why it beats several better-known names on code ownership. Read the tiers carefully before you commit, because the headline number does not buy the platform most buyers want: iOS output starts at the USD 200 tier, while USD 20 covers Expo, Android and web only. If the reason you are here is an iPhone app, that is a ten-fold difference between the plan you were looking at and the plan you need. The top tier compounds the problem: its credit ladder has no published prices above its first step, so the ceiling is a conversation rather than a number. Judged on what it makes rather than how it sells, it is a solid tool for a narrow job.

  Best for founders who want a credible native app prototype on a real device the same afternoon · From USD 20 / month, billed monthly · Last verified 20 August 2026 · Español

  3.7 / 53 ratings
• 

  Zite Review (2026)

  Buy for a website, skip for an application. Zite is exceptional value if what you actually need is a handsome, fast, well-marked-up site with a form on it: publishing takes minutes, the on-page SEO defaults are better than most full-stack builders manage, and the entry price is among the lowest here at USD 15. It scored 3.8 on our SEO and GEO axis, ahead of Replit and Base44, which tells you where its engineering attention has gone. The ceiling arrives fast though. Once you need user roles or a relational model, Zite is not the tool.

  Best for marketing sites, landing pages and simple lead-capture apps that need to look good and rank · From USD 15 / month, billed annually · Last verified 20 August 2026 · Español

  3.6 / 513 ratings
• 

  a0.dev Review (2026)

  The deployment story is the reason to look at it, and we had understated it. a0.dev does full one-click publishing to the App Store and Google Play, including creating the App Store Connect listing and uploading the build for you, plus over-the-air updates afterwards. For a solo builder who has never fought an Apple submission, that is worth more than another point of design polish. What you should not assume is a backend. There is no a0 datastore: the vendor describes the backend as powered by Convex or Supabase, which you sign up for separately, so budget a second vendor from day one. Several cells on this page previously claimed capabilities the vendor does not publish, and they now say so plainly: authentication and GitHub sync are not published by the vendor as of August 2026, meaning we could not verify them either way rather than that they are absent. Custom domain is a real Pro feature, listed as custom domain and no a0 branding. One published weakness to plan around: Android in-app payments are explicitly not supported yet. There is still no public API and no MCP server.

  Best for rapid mobile prototyping where the loop between an idea and seeing it matters more than anything else · From USD 20 / month, billed monthly · Last verified 20 August 2026 · Español

  3.6 / 53 ratings
• 

  Bravo Studio Review (2026)

  A specialist that does one thing beautifully, and a second thing we had missed. If you have a finished Figma file and an API, Bravo turns them into a native app that looks exactly like the design, and design at 4.4 reflects that fairly. The part our previous verdict denied is Bravo MCP, shipped in Bravo 4.0 as a beta inside the Solo plan: in the vendor's words it "plugs into Claude, ChatGPT, Cursor, or any MCP-compatible AI client and lets you build native apps from inside the AI conversation you already use", with React Native output, a Convex backend included and realtime sync from Figma. So "no backend, bring your own API" is wrong on that path. Auth and payments are further along than we said too: OAuth 2.0 login, custom authentication and Firebase email and phone login are Solo platform features, with Stripe and in-app purchases as an add-on. The real ceiling is one we never mentioned and it is the honest explanation of the 2.6 scalability mark: 15 screens per app on Starter, 30 on Solo. That is what limits Bravo to small applications, not the absence of features.

  Best for designers with a finished figma prototype and an api to plug into it · From USD 22 / month, billed monthly · Last verified 20 August 2026 · Español

  3.6 / 53 ratings
• 

  Emergent Review (2026)

  Interesting, not yet dependable. Emergent has the most ambitious agent architecture in this index: give it a real brief and it produces a task decomposition that reads like something a competent tech lead would write, then works through it across many files without asking permission every thirty seconds. When it lands, it lands big. Our panel gave it 4.2 on agent performance, among the highest here. The trouble is variance. The same brief run twice produced materially different quality, and the failure mode is quiet: plausible code that does not work, discovered later than you would like.

  Best for developers who want to hand off a well-specified multi-file change and review the result rather than watch it happen · From USD 17 / month, billed annually · Last verified 20 August 2026 · Español

  3.6 / 519 ratings
• 

  Manus Review (2026)

  Impressive, expensive and not really a builder. Manus plans and executes long chains of work in its own sandbox, and it scored 4.3 on agent performance because it genuinely completes tasks that stump single-purpose tools. As an app platform it is weak: no project structure to return to, scalability at 3.0, and credit consumption that made it the least predictable cost in our tests. Use it for a one-off build or a task that spans research, data and code. Do not use it as your product surface.

  Best for one-off builds, research-plus-build tasks and automations that span more than one tool · From USD 20 / month, billed monthly · Last verified 20 August 2026 · Español

  3.6 / 518 ratings
• 

  Draftbit Review (2026)

  A better-equipped product than our previous verdict described. Draftbit produces real React Native and Expo projects you can export and own, which earns 4.2 on code ownership, and the visual editor is pleasant. The current feature list is where we were most out of date: Claude Code and OpenAI Codex integration, bring-your-own AI across Claude, ChatGPT, OpenAI, Anthropic and OpenRouter with the vendor stating that Draftbit credits are not required, a built-in Supabase backend on every tier including Free, sandbox concurrency, multi-agent execution, custom MCP server integrations, and web and website output alongside mobile. Read the tier gates carefully: full code editing and export start at Standard, export to GitHub starts at Pro, and the custom MCP server integrations on Pro mean Draftbit consuming your MCP servers, not publishing one. Draftbit ships no MCP server of its own and no platform API, so its build lifecycle is still not agent-callable from outside, which is what the 3.4 API mark measures.

  Best for react native developers who want a visual head start, their own ai keys and a backend that is already there · From USD 12 / month, billed monthly · Last verified 20 August 2026 · Español

  3.5 / 55 ratings
• 

  Bubble Review (2026)

  Still the most capable platform here, still the hardest to leave. Bubble scores 4.5 on integrations because thirteen years of plugins and a large contractor market genuinely solve problems the AI-native tools cannot, and its native mobile story is better than we published: the vendor's own FAQ says "In Bubble, native mobile apps run on React Native", with store publishing, build submissions and unlimited over-the-air updates, so calling it a wrapper was inaccurate. The Data and Workflow APIs are real and useful. What does not exist is an MCP server: the vendor's March 2026 founder AMA says it is "on our radar, but not on the roadmap yet", the editor API is not open either, and every "Bubble MCP" you will find is a community project on forum.bubble.io. The other costs are unchanged: no code export at all, one of the weakest performance profiles in the index at 3.0 on speed, and workload pricing that surprises people in month three, though storage at least is published flatly at USD 3 per 100 GB a month. Choose it deliberately, with an exit you have accepted you do not have.

  Best for complex operational products where the plugin ecosystem and the contractor pool matter more than speed · From USD 59 / month, billed annually · Last verified 20 August 2026 · Español

  3.5 / 542 ratings
• 

  Trickle Review (2026)

  Buy on looks, and know what you are buying. Trickle produces the most tasteful output of the lightweight group and scored 4.3 on design, above every full-stack builder except Lovable and Totalum. If your job is a microsite, a campaign page or an interactive form that has to look considered, this is a genuinely good tool at a genuinely low price. It is also, by our measurements, the least capable backend here: scalability 2.9 and API and MCP 2.5, the latter the bottom of the index. Nothing about that is a bug; it is the product.

  Best for designers and marketers building visually distinctive pages, microsites and interactive forms · From USD 17 / month, billed monthly · Last verified 20 August 2026 · Español

  3.5 / 59 ratings
• 

  Rocket.new Review (2026)

  Buy only if mobile is the point. Rocket.new is the clearest answer in this index to a specific question: what if the output needs to be an app store app rather than a website? It generates Flutter and React Native targets from the same brief, the screens look genuinely native rather than a web page in a shell, and for a team with no mobile engineers that is a real shortcut. Everything else is mid-table or below. Reliability is shaky, the SEO axis score of 3.0 is among the weakest here, and the credit model prices exploration higher than it should.

  Best for teams who need a credible mobile app prototype and do not have mobile engineers · From USD 25 / month, billed monthly · Last verified 6 September 2026 · Español

  3.5 / 58 ratings
• 

  AppMaster Review (2026)

  The most unusual proposition in the index, and the one whose headline promise we got wrong. AppMaster does not interpret your project at runtime, it compiles it, and we used to recommend it on the strength of source you could download and walk away with. The vendor's own pricing FAQ says otherwise, verbatim: "Downloading source code available only for accounts with Enterprise Plan. Contact us to request Enterprise plan." Startup and Startup+ get database and file-storage export only. Business, from USD 955 a month, adds backend binaries and Docker images. Source is quote-only Enterprise. For every buyer below a custom contract, the thing we recommended it for does not exist, which is what 2.4 on code ownership now reflects. What survives is real: compiled Go services behind a 4.3 scalability mark, 4.3 on speed, and generated apps that come with documented REST APIs. Note which API we rate, though. The generated apps expose documented REST APIs; AppMaster itself publishes no API for driving the platform and no MCP server, and the API and MCP axis rates the platform, not its output, which is why it sits at 2.6. The interface is dense and dated, design output is plain, and the entry paid tier starts at USD 175 a month billed yearly, or USD 195 billed monthly, the highest entry price in the index.

  Best for teams on a business or enterprise contract who want compiled services they can run on their own infrastructure · From from USD 175 / month, billed yearly · Last verified 20 August 2026 · Español

  3.4 / 53 ratings
• 

  Knack Review (2026)

  Unfashionable, quietly dependable, and priced in a way nothing else here is. Knack has been building database applications for fifteen years and it shows in both directions: the data modelling and role handling are solid and it stayed up through every one of our tests at 4.2 on reliability, but the interfaces look like 2016 and the AI assistance is the weakest in this cohort at 2.8. The fact we previously inverted is the one most likely to decide the purchase. Every Knack tier lists "Unlimited Users / Roles", and the vendor titles its own pricing page "Knack Pricing | Unlimited Users Allows You to Scale Without Fees". Knack charges on records, storage and features, and never on seats, on every tier. In an index where per-seat pricing is the norm, that is the single most decision-relevant thing about it. Two more corrections: Ecommerce Payments is a first-party Pro-tier feature, so it is not as unsuited to anything a customer touches as we implied, and custom domains are tier gated at one on Pro and three on Corporate, with none on the entry tier, so "paid plans" was never a useful qualifier when every Knack plan is paid. There is still no free tier, only a trial.

  Best for membership registers, applications and case management for organisations without engineers · From USD 59 / month, billed monthly · Last verified 20 August 2026 · Español

  3.4 / 510 ratings
• 

  Adalo Review (2026)

  The kindest on-ramp in the index and no longer the right answer for most people. Adalo remains easy to learn and will genuinely get an app into the stores, and two facts in its favour deserve more prominence than we gave them: the database is a hosted PostgreSQL database, available on every tier including Free, and the vendor promises no usage or token-based charges on any tier, which is a real differentiator in a market that meters credits. Against that, performance is poor at 3.0, scalability at 2.8 is among the lowest here, there is no code export, and the App and Collections API is a Team-plan feature with no MCP server anywhere, so a bare yes on programmable access would mislead anyone pricing the cheaper plans. FlutterFlow now does more for less commitment.

  Best for first-time builders who want a phone app in the stores without learning a framework · From USD 36 / month, billed annually · Last verified 20 August 2026 · Español

  3.2 / 515 ratings
• 

  Thunkable Review (2026)

  A teaching tool that is priced and gated like a production platform, and the gating is the story. You cannot publish a live app on the entry paid tier: live publishing starts at Builder and covers exactly one app, and unlimited live apps require Advanced. That makes the Free and Accelerator tiers evaluation-only, which is not how they read at a glance. Worse for anyone thinking long term, the vendor FAQ states that "you must maintain an active subscription for your published apps to remain active". Your shipped app is a rental. That is materially worse ownership than any code-exporting builder here and it is the whole explanation for the lowest code-ownership score in the index. Two more line items people miss: you still need paid Google and Apple developer licences, which are separate from the Thunkable subscription, and AI tokens expire at the end of every billing cycle rather than rolling over. As a way to teach app logic through drag-and-drop blocks, Thunkable remains genuinely good, and the classroom is where it belongs.

  Best for education, prototypes and simple utility apps built by people learning to build at all · From USD 18 / month, billed yearly · Last verified 20 August 2026 · Español

  3.1 / 511 ratings