Llama Coder
Llama Coder is an open-source app generator under MIT that its README calls an open-source Claude Artifacts: describe a small app in one prompt and it writes the code and runs a live preview in your browser. It uses Meta's Llama 3.1 405B through Together AI for the model and renders the preview in a sandboxed frame. It is a Next.js app you run with a Together AI key and a Postgres database.
What does Llama Coder do?
Llama Coder's README calls it "an open source Claude Artifacts" that can "generate small apps with one prompt", powered by Llama on Together AI. The model is "Llama 3.1 405B from Meta", served by "Together AI for LLM inference." The app previews what it generates with "esbuild-wasm + esm.sh for the in-browser preview renderer (runs in a sandboxed iframe)."
It is built as a Next.js app with the App Router and Tailwind. Its own data lives in PostgreSQL on Neon through Prisma, screenshot uploads go to an S3 bucket, and Braintrust observability and Plausible analytics are optional. The README does not describe a database or deployment for the apps it generates; the output is a single small app with its preview.
Key facts
- Licence: MIT, copyright Hassan El Mghari, with no added conditions.
- Generates small apps from a single prompt, in the manner of Claude Artifacts.
- Model: Llama 3.1 405B from Meta, through Together AI. The README's tagline says Llama 3, so it is not consistent with itself.
- Preview: esbuild-wasm and esm.sh in a sandboxed iframe, in the browser.
- Built with Next.js, the App Router and Tailwind; its own data in Postgres through Prisma.
- Optional: S3 for screenshot uploads, Braintrust for observability, Plausible for analytics.
- Run: clone, set a Together AI key and a database URL, npm install and npm run dev.
- No GitHub releases; last push when read 2026-09-15; 7,131 stars on 2026-09-17.
What does it replace, and where does it fall short?
Llama Coder replaces v0 and Bolt for the quick end of the job: one prompt, one small working app, previewed instantly, built on an open model and a codebase you can read and change in an afternoon. It is the pick for small one-prompt apps on the open-source alternatives to v0 page.
Where it falls short: it builds small, single apps with no backend, database or deployment for them, and it is tied to Together AI for inference. It is a template to learn from and extend as much as a product. For full-stack apps, Chef; for a desktop builder, Dyad.
How does Llama Coder run?
Clone the repository, create an .env file with a Together AI API key and a Postgres database URL, then npm install and npm run dev. Add S3 details if you want screenshot uploads. A hosted version runs at llamacoder.io; the README describes no paid tier.
Who is Llama Coder for?
A developer who wants a small, readable prompt-to-app generator to run, study or extend, on an open model. Someone building a full product should start with Chef or Dyad.
What limits does the README state?
None stated beyond its own scope, "generate small apps with one prompt".
Questions people ask
Is Llama Coder open source?
Yes. It is MIT licensed with no added conditions, so you may run it, change it and build on it. A hosted version runs at llamacoder.io, and the README describes no paid tier.
Which model does Llama Coder use?
Llama 3.1 405B from Meta, served through Together AI, by its README; its tagline still says Llama 3. You need a Together AI API key. The README does not describe other providers or local models.
What kind of apps can Llama Coder build?
Small apps from a single prompt, previewed in a sandboxed frame in your browser, in the manner of Claude Artifacts. The README does not describe databases, sign-in or deployment for the generated apps, so it suits prototypes and single-screen tools.
How does Llama Coder compare with v0?
Llama Coder is a small, MIT-licensed generator on an open model that you can run and modify; v0 is Vercel's hosted platform for designing and scaling full-stack web apps. The open-source alternatives to v0 page compares it with Onlook, Dyad and Chef.
What does it take to run Llama Coder myself?
A Together AI API key and a Postgres database URL in an .env file, then npm install and npm run dev. S3 settings are needed only for screenshot uploads, and Braintrust and Plausible are optional.
Sources
- Llama Coder README and MIT LICENSE: github.com/Nutlope/llamacoder, read 2026-09-17.
Compared with the others
On the open-source alternatives to v0 page, Llama Coder is the pick for small apps from one prompt. MIT, generating a small app from a single prompt with Llama 3.1 405B through Together AI and previewing it in a sandboxed frame.
Also on that page: Onlook for editing the result visually, Dyad for on your own computer, with your own keys, Chef for a database and sign-in generated too.
On the open-source alternatives to Bolt page, Llama Coder is the pick for small apps from one prompt. MIT, generating a small app from a single prompt with Llama 3.1 405B through Together AI and previewing it in a sandboxed frame.
Also on that page: Chef for a database and sign-in generated too, Dyad for on your own computer, with your own keys, Onlook for editing the result visually.
On the open-source alternatives to Lovable page, Llama Coder is the pick for small apps from one prompt. MIT, generating a small app from a single prompt with Llama 3.1 405B through Together AI and previewing it in a sandboxed frame.
Also on that page: Dyad for on your own computer, with your own keys, Onlook for editing the result visually, Chef for a database and sign-in generated too.
More ai app builders
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.
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.
Added September 17, 2026. Every claim above comes from the project's README, LICENSE or model card, read on September 17, 2026, or from the GitHub API on the date shown in the panel. Found an error? Write to reveneau@licheo.com and it is fixed in the next weekly pass. Repository: github.com/Nutlope/llamacoder.
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.