Press release:

Arabic.AI, a regional leader in Arabic artificial intelligence and enterprise technology, announced the launch of HELM Arabic Enterprise in collaboration with Stanford University’s Center for Research on Foundation Models (CRFM). The initiative is designed to strengthen how organisations evaluate Arabic large language models (LLMs) for enterprise use. 

Stanford’s CRFM is known for creating the HELM (Holistic Evaluation of Language Models) framework, which has set a global standard for transparent and reproducible model evaluation. Building on that foundation, HELM Arabic Enterprise introduces a structured benchmark that gives the Arabic AI ecosystem a practical, shared reference for comparing model behavior and supporting more consistent evaluation practices. 

HELM Arabic Enterprise evaluates models across six enterprise-focused tasks spanning content generation, financial reasoning, and legal question answering. The benchmark is designed to measure how reliably Arabic LLMs perform in professional and institutional use cases, particularly in regulated environments. As with all HELM benchmarks, prompts, responses, metrics, and scores are transparent and reproducible through the open-source HELM framework.

For Arabic.AI, the collaboration aligns with its long-term goal of advancing Arabic-first AI while contributing tools that are useful to the broader research and enterprise community. The release of HELM Arabic Enterprise provides teams with a common baseline they can use for internal assessment, vendor comparison, and ongoing model oversight. Arabic.AI and Stanford’s CRFM view this as an important step toward more mature benchmarking infrastructure for Arabic enterprise AI.  

“Arabic enterprise AI needs evaluation framework that is rigorous, open, and directly tied to real business workflows,” said Nour Al Hassan, CEO of Arabic.AI. “HELM Arabic Enterprise gives the ecosystem a shared benchmark to measure progress and reliability with clarity and confidence.”