The open-source alternatives to Lovable are Dyad, Onlook, Chef and Llama Coder. Dyad runs the whole prompt-to-app loop on your own Mac or Windows machine with your own API keys and no sign-up. Onlook adds visual editing on the live page for Next.js and Tailwind projects. Chef, forked from Bolt's open code, generates a database, sign-in and background jobs along with the interface. Llama Coder builds small single-prompt apps on an open model. A builder switches to own the code and pay the model provider directly.
4 of 4 carry an OSI licence. 4 of 4 run on your own machines. Figures read from GitHub on September 17, 2026.
One project per need, with the reason from its own README and the one thing to know before you choose it.
A desktop app for Mac and Windows whose README compares it to Lovable, v0 and Bolt running locally, with no sign-up and your own API keys.
Know firstIts README does not name the models, framework or deployment targets, and there is no Linux build.
Apache-2.0 visual-first editor for Next.js and Tailwind, with a Figma-like interface, layers, branching and checkpoints, synced to real code.
Know firstNext.js and Tailwind only for now, and running it yourself needs Bun, Docker, Supabase and OpenRouter.
Apache-2.0, forked from bolt.diy, building full-stack apps with a database, authentication, file uploads and background jobs on Convex.
Know firstEvery app it builds runs on Convex, and self-hosting uses Convex's hosted control plane.
MIT, generating a small app from a single prompt with Llama 3.1 405B through Together AI and previewing it in a sandboxed frame.
Know firstSmall single apps with no backend or deployment, tied to Together AI for inference.
The facts that decide most choices. Each project's page carries the full panel.
| Project | Best for | Openness | Licence | Runs as | Own machines | Stars |
|---|---|---|---|---|---|---|
| Dyad | On your own computer, with your own keys | OSI | See LICENSE | Desktop app | Yes | 21,576 |
| Onlook | Editing the result visually | OSI | Apache-2.0 | Hosted app, Run locally (Bun, Docker, Supabase) | Yes | 26,756 |
| Chef | A database and sign-in generated too | OSI | Apache-2.0 | Hosted app, Run locally (pnpm, Convex) | Yes | 4,610 |
| Llama Coder | Small apps from one prompt | OSI | MIT | Web app (Next.js) | Yes | 7,131 |
Licence and star figures read from GitHub on September 17, 2026. Stars are shown as a dated fact and were not used to rank this page.
A local, open-source AI app builder for Mac and Windows that its README compares to Lovable, v0 and Bolt running on your own machine, with your own API keys and no sign-up, under Apache-2.0 apart from a fair-source pro directory.
A desktop app for Mac and Windows whose README compares it to Lovable, v0 and Bolt running locally, with no sign-up and your own API keys.
Know firstIts README does not name the models, framework or deployment targets, and there is no Linux build.
An Apache-2.0 visual-first code editor for building Next.js and Tailwind sites and prototypes with AI, editing the page directly in the browser with a Figma-like interface while the code stays in sync.
Apache-2.0 visual-first editor for Next.js and Tailwind, with a Figma-like interface, layers, branching and checkpoints, synced to real code.
Know firstNext.js and Tailwind only for now, and running it yourself needs Bun, Docker, Supabase and OpenRouter.
Convex's Apache-2.0 AI app builder, a fork of bolt.diy, that builds full-stack web apps with a built-in database, authentication, file uploads, real-time interfaces and background jobs on the Convex backend.
Apache-2.0, forked from bolt.diy, building full-stack apps with a database, authentication, file uploads and background jobs on Convex.
Know firstEvery app it builds runs on Convex, and self-hosting uses Convex's hosted control plane.
An MIT open-source take on Claude Artifacts that generates small apps from one prompt with Llama 3.1 405B on Together AI and previews them in a sandboxed in-browser renderer.
MIT, generating a small app from a single prompt with Llama 3.1 405B through Together AI and previewing it in a sandboxed frame.
Know firstSmall single apps with no backend or deployment, tied to Together AI for inference.
The questions that settle it, in the order they usually come up.
Dyad is a desktop app with your own keys. Onlook and Chef can be self-hosted or used hosted. Llama Coder runs locally with a Together AI key.
Chef generates a database, sign-in, file uploads and background jobs. The others generate front-end code.
Onlook edits the live page with a Figma-like interface while keeping the code in sync.
Lovable, in the vendor's own words, is "a full-stack AI development platform for building, iterating on, and deploying web applications using natural language, with real code, security, and enterprise governance", by its own description on its documentation site. lovable.dev. The cases below are where that still wins.
Stars, licence, last commit and last release are read from the GitHub API by a script every week and carry the date they were read. Nobody types them.
What each project does is written from its own README and LICENSE, read in full on the date shown, and phrased as the project's claim. The sources are listed at the end.
That a project is faster, better or cheaper than Lovable. A price for anyone. A user count. A roadmap. If a fact is not on this page, it was not in the source.
Dyad, whose README makes the comparison itself: an app builder like Lovable, v0 and Bolt, running on your machine with your own keys. Onlook if you want to edit the result visually, Chef if you need a backend generated too, Llama Coder for small apps on an open model.
Yes. Dyad is a desktop app for Mac and Windows with no sign-up, using your own API keys. Llama Coder is a Next.js app you run locally with a Together AI key. Onlook and Chef can be self-hosted, though both also offer hosted versions.
Chef, from Convex. Its README says it builds full-stack web apps with a built-in database, zero-configuration authentication, file uploads, real-time interfaces and background workflows. The trade is that those apps run on Convex.
Yes. Each generates code you hold: Onlook works on any Next.js and Tailwind project and shows the exact line behind any element, Chef writes a project on a public template, and Dyad runs locally so the files never leave your machine.
The software is free under each licence. You pay whichever model provider you bring a key for. Dyad's pro directory is under a licence that forbids competing commercial use until it converts to Apache-2.0 two years after each release; the rest are Apache-2.0 or MIT.
Running one of these inside your own environment, with your own data and your own security rules, is the kind of work Reveneau does. Read how a forward deployed engagement works.