ALLaM 7B
Saudi Arabia's national Arabic LLM: a 7B open-weight model for Arabic understanding and generation.
About
ALLaM 7B is the generative Arabic large language model developed by Saudi Arabia's SDAIA (Saudi Data and Artificial Intelligence Authority), listed publicly on Hugging Face in March 2025 and re-released as ALLaM-7B-Instruct-preview under HUMAIN in July 2025. It was developed with IBM, initially served through IBM's watsonx platform and later on Microsoft Azure, as part of the Kingdom's Vision 2030 push to build domestic AI capacity.
The model targets Arabic text understanding, summarization, and content generation, trained on a large and varied Arabic dataset. Weights are published under Apache 2.0, which makes it one of the few permissively licensed Arabic-first models at this scale. SDAIA describes it as among the leading generative models for Arabic, but no public benchmark tables or independent evaluations were found, so that positioning should be read as the provider's own claim rather than a measured result.
Specs
Context window
Pricing
Free quota
API access
Weights
Released
Benchmarks
No published benchmarks found; SDAIA's 'leading Arabic model' positioning is a provider claim without a public evaluation table
Provider-reported unless a source says otherwise. Benchmarks are marketing-adjacent by nature — treat them as directional.
Use cases
Trust score — 6/10
Scored 6/10 (metrics tier, metrics-v1). Evidence: real national-lab release (SDAIA, Mar 2025; HUMAIN instruct preview Jul 2025) with Apache-2.0 weights on Hugging Face; distribution via watsonx and Azure corroborated by Saudi press. Open weights. Strengths: rare permissively licensed Arabic-first model, state-backed maintenance. Docked for: (1) no public benchmark table or independent evaluation found; (2) modest community traction (~11K HF downloads) and thin technical documentation.
How the trust score is built
Full methodology →- Adoption · stars/downloads, log-scaled vs peers
- 5.0/10
- Maintenance · recency, releases, issue load
- 6.0/10
- Openness · license + docs
- 8.0/10
- Manual review · sanity read + red-flag check
- 6.0/10
- · SDAIA national Arabic LLM; public HF listing Mar 2025, HUMAIN instruct preview Jul 2025
- · Developed with IBM (watsonx), also on Azure; ~11K HF downloads
- · Apache-2.0 weights — rare permissively licensed Arabic-first model
- · No public benchmark table; thin technical documentation