Bayanat Labs Arabic AI Model Tracker
A register of Arabic-centric and Arabic-capable AI model releases from 20 May 2023 to 7 October 2026: text LLMs, vision-language and OCR models, speech recognition, text to speech, embeddings, translation and dialect identification. For each release it records the lab, the date, the modality, the sizes the lab states, the Arabic varieties the lab names (quoted, with a link), the benchmark scores the lab reports, the licence as stated, and whether the weights are open, gated, hosted, announced without release or retired. Each row links to its source, with the retrieval time and the SHA-256 of the archived page. The tracker records facts as the labs state them; it does not score, rank or compare models. Version v0.1.0 is a backfill from the labs' own pages, not a complete census. A crawler lists new candidates from Hugging Face, arXiv and the OpenRouter models list for review, and nothing is added automatically.
| Licence | CC BY 4.0 (register, exports, charts and documentation); Apache-2.0 (crawler and build code); quotations stay with their authors |
|---|---|
| Version | v0.1.0 · Register and exports v0.1.0, as of 19 October 2026; the code is tagged on GitHub and not on PyPI yet. |
| Published | |
| Updated | |
| Who it is for | Teams that choose, compare or follow Arabic-capable models and need each release's lab, date, sizes, licence and weights status with a source link, or a feed of new releases. |
| Level | Beginner |
Key numbers
Counted by script from the released files. Each key number is read from summary.json in the Hugging Face dataset and pinned to its SHA-256; MANIFEST.json lists every file with its SHA-256.
Downloads and identifiers
| Where | What |
|---|---|
| Hugging Face | Dataset on Hugging Face |
| bayanatlabs.com | Browse and filter the release table |
| Hugging Face | Atom feed of new releases |
| huggingface.co | Releases table (CSV) |
| Hugging Face | latest.json: headline counts and recent releases |
| GitHub | Code and inclusion rules on GitHub |
| GitHub | Inclusion rules |
Get started
Load the tables with the datasets library, or install the crawler and export builder from GitHub at the v0.1.0 tag to validate the register and rebuild the exports; the package is not on PyPI yet.
# Python
from datasets import load_dataset
releases = load_dataset("Bayanat-Labs/arabic-ai-model-tracker", "releases", split="train")
scores = load_dataset("Bayanat-Labs/arabic-ai-model-tracker", "benchmarks_reported", split="train")
# Crawler and export builder
git clone --branch v0.1.0 https://github.com/Bayanat-Labs/arabic-ai-model-tracker
cd arabic-ai-model-tracker
pip install .
arabic-model-tracker validate --as-of 2026-10-19What is released
| Part | Licence | In this release |
|---|---|---|
| Register: releases, checkpoints and lab-reported benchmark tables (CSV and Parquet), charts, Atom feed, latest.json and the candidate queue | CC BY 4.0 | Yes |
| Crawler, candidate scorer, checkpoint snapshot and export builder | Apache-2.0 | Yes |
| Short quotations from model cards, papers and announcements | Kept with their source links to show what a lab states; they remain the property of their authors | Yes |
Attribution: Arabic AI Model Tracker by Bayanat Labs, CC BY 4.0. https://huggingface.co/datasets/Bayanat-Labs/arabic-ai-model-tracker
Intended uses
- Shortlist Arabic-capable models by modality, licence and weights status, then read each lab's own source.
- Follow new Arabic model releases through the Atom feed or latest.json.
- Count and chart Arabic model releases over time, citing the version and as-of date.
Out of scope
- Ranking or comparing model quality: scores are copied as each lab reports them, on different benchmarks and settings.
- Licence advice: the licence is recorded as stated; read the licence itself before you use a model.
Rights at a glance
Training, display, redistribution and synthetic-voice use are separate permissions.
| Use | Status | Note |
|---|---|---|
| Training | Allowed | Register and exports under CC BY 4.0, with attribution. |
| Display | Allowed | CC BY 4.0, with attribution; keep each quotation with its source link. |
| Redistribution | Allowed | Register under CC BY 4.0; code under Apache-2.0. Model names, licence names and lab-reported scores are cited facts with source links. |
| Synthetic voice | Not applicable | Metadata only; no audio and no model weights. |
Limitations
- Version v0.1.0 is a backfill from the labs' own pages, not a complete census: a quarter with fewer rows may have fewer recorded releases rather than fewer releases.
- A release is recorded only when the lab names Arabic in its own card, paper or announcement; a language tag in repository metadata is not enough.
- When a release has no dated announcement or paper, its date is the creation date of its Hugging Face repository, which can be earlier than public availability.
- Install the crawler from GitHub for now; it is not on PyPI yet.
Sources and licences
| Source | Licence | Redistributed here |
|---|---|---|
| Model cards and repository metadata on the Hugging Face Hub | Facts and short quotations with source links; each card under its own terms | No |
| Labs' papers on arXiv and their own announcements | Facts and short quotations with source links; arXiv API metadata under CC0 1.0 | No |
| OpenRouter public models list | Used to find candidates for review; nothing copied | No |
Each row is written by Bayanat Labs from the releasing lab's own model card, paper or announcement. Source pages are archived for verification; the register publishes their URLs, retrieval times and SHA-256 hashes, not the pages. Source details
How to cite
Cite as:
Bayanat Labs (2026). Bayanat Labs Arabic AI Model Tracker, version 0.1.0 (as of 2026-10-19). Bayanat Labs. https://huggingface.co/datasets/Bayanat-Labs/arabic-ai-model-tracker
No DOI. Cite the URL in the citation above and the date you read it.
@misc{bayanatlabs2026arabicaimodeltracker,
author = {{Bayanat Labs}},
title = {Bayanat Labs Arabic AI Model Tracker, version 0.1.0 (as of 2026-10-19)},
year = {2026},
month = oct,
howpublished = {\url{https://huggingface.co/datasets/Bayanat-Labs/arabic-ai-model-tracker}}
}
Changelog
| Date | Version | Change |
|---|---|---|
| v0.1.0 | Register, exports, charts, Atom feed and latest.json on Hugging Face; crawler, inclusion rules and export builder on GitHub; the filterable release table on this site. |
Corrections and removal requests
If something in this release should not be there, or is wrong, email hello@bayanatlabs.com. We handle removal requests within 30 days, and platform requests within 24 hours. How removal requests work
Related product: Bayanat Bench