AraBERTv2
Arabic BERT pretrained on 77GB of text, the standard Arabic NLU baseline.
About
AraBERTv2 is the second-generation Arabic BERT from the American University of Beirut's MIND Lab (arXiv 2003.00104, 2020). It comes in base (136M) and large (371M) sizes, pretrained on 200M sentences / 77GB of Arabic text with Farasa-based pre-segmentation of prefixes and suffixes, a key adaptation for Arabic morphology. It superseded v1 with a much larger corpus.
AraBERT became the default baseline for Arabic NLU, evaluated on sentiment analysis, NER (ANERcorp), and question answering (Arabic-SQuAD, ARCD). Checkpoints are public on Hugging Face (aubmindlab) in PyTorch, TF2, and TF1 formats. Release date is era-level (2020).
Specs
Context window
Pricing
Free quota
API access
Weights
Released
Benchmarks
Provider-reported unless a source says otherwise. Benchmarks are marketing-adjacent by nature — treat them as directional.
Use cases
Trust score — 7/10
Scored 7/10 (metrics tier, metrics-v1). Evidence: Verified via vendor site/docs/GitHub and reputable coverage, wave 119; Trending score 68/100 from researcher assessment.