Live · credited

Arabic models, from the source

OALL is the official Arabic LLM scoreboard. Hugging Face is where the weights live. This page is a live window onto both. We do not invent OALL averages. We fetch public model cards and send you to the original board.

Open OALL (source) Run a GGUF locally

Source of truth. Rankings belong to OALL (Open Arabic LLM Leaderboard), run by 2A2I + TII + Hugging Face. Weights belong to their authors on Hugging Face. ABUZ8 is not affiliated with either.

OALL v2

Native Arabic tasks: ArabicMMLU, MadinahQA, AraTrust, ALRAGE, AlGhafa. Submit a model there if you want a score.

huggingface.co/spaces/OALL →

Hugging Face hub

Live trending models, datasets, Spaces, and GGUF files from the public API.

Stay on abuz8ai.com →

Local GGUF

Files you can run with Ollama, llama.cpp, or LM Studio. No API key.

Open the local tab →

Live Arabic-related checkpoints from Hugging Face search (ALLaM, Jais, Fanar, SILMA, AceGPT, Hala, Falcon-H1). Sorted by downloads. Each card opens the author's page.

Contacting Hugging Face through /api/hf-live…

Data: Hugging Face public API, proxied at /api/hf-live?kind=arabic. Download and like counts belong to each author. Official Arabic ranking: OALL, not this page.