Participate in SILMA AI's Annual Survey: The State of AI in MENA
العربية
We build Arabic AI Technologies to solve your problems
SILMA AI develops advanced Arabic AI technologies to empower individuals and enterprises in the region with sovereign technologies that are tailored to their needs
SILMA Large Language Models (LLM) are top ranked on Arabic Language benchmarks
Our Promise
Small Models. Big Performance
High-Performance Models
Our models outperform larger models in Arabic language tasks, despite being small and efficient
Cost-Effective Solution
Faster to run and less expensive to host, making it practical for business use-cases
Powered by Google Gemma
Built over Google's open-weight models, combining English capabilities with Arabic excellence
Local Hosting Capability
Unlike closed models, can be hosted locally in your data-center or private company cloud
Tailored for you
Our team can tailor the model to outperform any existing open-source model for your unique requirements
Our Models
LLMs - General Purpose

huggingface

silma-ai/SILMA-9B-Instruct-v1.0 · Hugging Face

We’re on a journey to advance and democratize artificial intelligence through open source and open science.

SILMA 9B Instruct v1.0 is revolutionizing Arabic language models with its exceptional performance. Despite its compact size of 9 billion parameters, it outperforms bigger models in terms of effectiveness and efficiency
LLMs - RAG

huggingface

silma-ai/SILMA-Kashif-2B-Instruct-v1.0 · Hugging Face

We’re on a journey to advance and democratize artificial intelligence through open source and open science.

SILMA Kashif 2B Instruct v1.0 is an advanced open-weight model designed for Retrieval-Augmented Generation (RAG) tasks, excelling in answering questions based on context in Arabic and English. It also has minor capabilities in Entity Extraction. Built on Google Gemma’s foundational models, it delivers top-tier performance within the 3-9 billion parameter range, as validated by the SILMA RAGQA Benchmark. The model supports a 12k context length
Embeddings

huggingface

Arabic Embedding Models - a silma-ai Collection

A collection of highly efficient and lightweight Arabic Embedding Models developed by SILMA AI, featuring innovative Matryoshka-based models.

SILMA Embedding STS v0.1 and SILMA Embedding Matryoshka v0.1 are highly efficient and lightweight Arabic Embedding Models featuring innovative Matryoshka-based models
Our Services
We offer the following services for Enterprise customers
1
Build
We build Arabic AI technologies
2
Consult
We collaborate with you to identify and implement innovative AI solutions that address your challenges
3
Tailor
We tailor our models to best fit your specific tasks by leveraging your data, ensuring optimal performance compared to any other model on the market
Struggling with business challenges?
Let us help you harness the power of AI to find innovative solutions. Reach out to us today via hello@silma.ai
Our Team
Mohamed El-Kamhawy
Co-Founder, Advisor
Karim Ouda
Co-Founder, CTO
Hany Ouda
Advisor
SILMA AI
19901, Dover, Delaware, United States
©SILMA.AI LLC - All Rights Reserved 2025