metaai·lightalo unofficial · independent

About this field guide

An independent, unofficial map of thirteen years of Meta AI research — how it was researched, how the Lab runs real models on a static site, what the badges mean, and how to tell us when we got something wrong.

119projects
12constellations
80lineage links
202languages
8SAM demo images
10MusicGen clips

01What this site is

metaai·lightalo is a field guide to everything Meta's AI labs have shipped — from the founding of FAIR in 2013 through PyTorch, Llama, Segment Anything, DINO, Seamless and the Muse era. It is written and maintained by one person, as a lightalo side project. It is not affiliated with, endorsed by, or sponsored by Meta, and nobody at Meta reviewed it.

It exists because this body of work is remarkable and surprisingly hard to see whole. The papers are scattered across arXiv, the code across hundreds of repositories, the announcements across years of blog posts — and much of it was released openly, for anyone to use. We wanted one place where a curious newcomer and a working researcher could both stand back and take in the shape of it. The tone is admiring on purpose; the facts are checked on purpose too.

02How it was built

The catalog began as a research sweep over primary sources: arXiv papers, GitHub repositories, ai.meta.com and about.fb.com announcements, plus reputable press for the facts that only exist in news coverage. Roughly 140 sourced entries were distilled into 119 projects, each placed in one of 12 constellations — The Muse Era, Language & LLMs, Vision, World Models & Embodied AI, Speech & Sound, Generative Media, AI for Science, Open Source Infra, Games & Strategy, Safety & Trust, On-Device & Silicon, and Reality Labs Research — and joined by 80 lineage links showing what grew out of what.

Every entry then went through an adversarial fact-check pass: a second reader was asked to find anything wrong rather than to confirm what was right. That pass corrected 33 claims — dates, model sizes, licence terms, who led what — before launch.

The Lab runs real models, entirely offline. Large models split into a heavy understanding half and a light interactive half, so we ran the heavy half once on a Raspberry Pi 5 and shipped the light half to your browser:

The whole site is static HTML, CSS and JavaScript: no build step, no frameworks, no analytics, no trackers, no cookies. Nothing you click here is sent anywhere.

03Sources & verification

Every star links to its primary sources. Open any project in the galaxy and the panel lists the paper, repository and announcement it was built from — they are the authority, not us. The catalog is a single file, /data/projects.json, which you are welcome to read, cite and build on.

The “verified 3 Sep 2026” stamp in our footers means that on that date every fact in the catalog was re-read against its cited source, and that every link on the Live Demos page was loaded and confirmed alive. It is a snapshot, not a promise: Meta ships quickly and links rot. When something changes, the stamp changes with it.

How to read the badges on each star:

open sourceWeights or code released under an open licence. closed / productA shipped product, API or research result without an open release. previewAnnounced, but not yet generally available. supersededHistorically important; replaced by a newer generation. archivedRepository frozen or project wound down.

04Licenses & attribution

The Lab's models belong to their creators; we use them within their licences and say so on every page:

ModelLicenceHow we use it
SAM (sam_vit_b)Apache 2.0Quantised decoder shipped to your browser; embeddings precomputed.
DINOv2 (vits14)Apache 2.0PCA features and patch pairs precomputed; an 8-bit ONNX export runs in your browser for your own photos.
OPT tokenizer (GPT-2 BPE)MITVocabulary and merge files served to your browser; tokenization runs locally.
NLLB-200 (600M)CC-BY-NC 4.0Outputs shown non-commercially, labelled as such.
MusicGen (small)CC-BY-NC 4.0Outputs shown non-commercially, labelled as such.

This site carries no advertising and sells nothing. The demo photographs are our own AI-generated images, made for these demos. Project names and marks belong to their owners; they appear only to identify the work being described, and no Meta logos or trade dress are used. The in-browser runtime is onnxruntime-web (MIT), served from this domain.

05Corrections

We would rather be corrected than be wrong. If a date is off, a licence is misstated, a lineage arrow points the wrong way, or a project that deserves a star is missing, please tell us — a link to the primary source makes it a five-minute fix.

✉ Send a correction rishabhdara@gmail.com · subject “metaai correction”

Fixes land in the catalog and refresh the verified date.

06Start here

Two ways in, depending on how much of this story you already know.

New to Meta AI?

  1. Read the Timeline — thirteen years in six acts, about ten minutes.
  2. Try Segment Anything — click an object, watch a real model cut it out.
  3. Then open the galaxy, press ▶ Play to watch thirteen years form star by star, and click a constellation name for its story.
  4. Test yourself with Which Meta model? — ten quick rounds.

Already know the field?

  1. Go straight to the galaxy: search, filter by era, or toggle open source only; press /Ctrl K anywhere to jump to any star or page.
  2. Untangle the family trees, then check By the numbers for every documented model size on one log-scale timeline.
  3. Deep-link any star — /#p=blt opens Byte Latent Transformer, and /p/blt/ is its shareable page; every project id works the same way. Or read the JSON directly.