What challenges did Core42 face before AI implementation?

Limited accessibility to state-of-the-art bilingual AI models for Arabic speakers Struggle in handling local government, public, and multilingual sectors Manual processes leading to inefficiencies in various industries Mediocre performance of existing AI models for Arabic text and data

What was the impact of the AI solution?

Performance improvement over other Arabic AI models Automation capabilities in patient management and customer service Scaled accessibility with Azure infrastructure Accelerated innovation in industries including education and banking

Rank#2,095of 3,107

Source

MicrosoftWent dark

JAIS: Transforming Arabic and English AI Solutions

Use case typeMultilingual communicationUpdated Jun 13, 2026

Microsoft, in collaboration with Core42, has launched JAIS, a bilingual (Arabic-English) GPT-3-derived large language model on Azure. Aimed at revolutionizing multiple industries including public and government sectors, JAIS optimizes communication, digitizes records, and enhances services in the region. It is already making significant inroads in other fields such as education, banking, healthcare, and media. The model employs 30 billion parameters and sophisticated techniques, setting benchmarks in Arabic-language tasks. Organizations can access JAIS on Azure Model catalog to explore its powerful AI applications.

Organization: Core42
Industry: Tech & Comms
Location: United Arab Emirates
Published: May 2024

Reported outcomes

Strategic outcomes

New product / capabilityLaunched a bilingual Arabic-English AI modelNew product / capabilityEnabled multiple AI applicationsCustomer experience & trustImproved services in public sectorsScale & capacityExpanded accessible AI via cloud infrastructure

Primary read

Use case focus

Showing 3 of 4

1language model
2multilingual AI
3chat AI

Limited accessibility to state-of-the-art bilingual AI models for Arabic speakers
Struggle in handling local government, public, and multilingual sectors
Manual processes leading to inefficiencies in various industries
Mediocre performance of existing AI models for Arabic text and data

Launched JAIS 30B Chat, a bilingual Arabic-English AI model
Implemented state-of-the-art architecture including ALiBi, SwiGLU
Deployed on scalable and secure Microsoft Azure AI
Supported multiple applications like chat, content creation, and data analytics
Offered accessibility through Models as a Service (MaaS) on Azure

Technologies

Azure AI

Performance improvement over other Arabic AI models
Automation capabilities in patient management and customer service
Scaled accessibility with Azure infrastructure
Accelerated innovation in industries including education and banking

Architecture

JAIS uses sophisticated AI architecture employing ALiBi positional embeddings and SwiGLU non-linearity. It efficiently processes Arabic texts with fewer tokens leveraging a custom tokenizer.

Implementation partners1

G42

Sources & evidence1

Went darkLost public footprint

The cited source is no longer reachable and the organization has no newer case. Not a claim the system was discontinued.