The Arabic LLM Company

Arabic AI Technologies to solve your business problems

We build, consult, and tailor open-source models to meet your unique needs

Our Services

Our Models

Who We Are

We are committed to advancing AI technologies that offer enterprises throughout the Arabian region sovereign, tailored solutions designed to meet their unique needs

Our Services

Your path to excellence

We offer the following services for Enterprise customers

Tailoring

We tailor our models to fit your specific tasks by leveraging your data, ensuring optimal performance compared to any other model on the market

class Sampling(layers.Layer):

"""Uses (mean, log_var) to sample z, the vector encoding a digit."""

def call(self, inputs):

mean, log_var = inputs

batch = tf.shape(mean)[0]

dim = tf.shape(mean)[1]

return mean + tf.exp(0.5 * log_var) * epsilon

class Sampling(layers.Layer):

"""Uses (mean, log_var) to sample z, the vector encoding a digit."""

def call(self, inputs):

mean, log_var = inputs

batch = tf.shape(mean)[0]

dim = tf.shape(mean)[1]

return mean + tf.exp(0.5 * log_var) * epsilon

class Sampling(layers.Layer):

"""Uses (mean, log_var) to sample z, the vector encoding a digit."""

def call(self, inputs):

mean, log_var = inputs

batch = tf.shape(mean)[0]

dim = tf.shape(mean)[1]

return mean + tf.exp(0.5 * log_var) * epsilon

Building

We build Arabic AI technologies to help you reach your goals

Security

Efficiency

Speed

Accuracy

Status:

Updating:

Security

Efficiency

Speed

Accuracy

Status:

Updating:

Security

Efficiency

Speed

Accuracy

Status:

Updating:

Consulting

We collaborate with you to identify and implement innovative AI solutions that address your challenges

Our Models

Small Models. Big Performance

Small yet powerful, our Arabic LLMs—built on Google Gemma—outperform larger models in Arabic tasks, offering cost-effective, high-performance solutions with local hosting capabilities for business use

Ask me something..

General Purpose LLMs

SILMA 9B Instruct v1.0

A 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

Ask me something..

RAG-optimized LLMs

SILMA Kashif 2B Instruct v1.0

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. Kashif delivers top-tier performance within the 3-9 billion parameter range, as validated by the SILMA RAGQA Benchmark. The model supports up to 12k context length

Ask me something..

SILMA Embedding

SILMA Embedding Matryoshka v0.1

SILMA Matryoshka is a highly efficient and lightweight Arabic Embedding Models featuring innovative Matryoshka-based models

Our Benchmarks

Arabic-focused Benchmarks

The Arabic Broad Leaderboard (ABL) is a next-generation benchmarking platform that offers advanced visualizations, performance analysis, model skill breakdowns, speed comparisons, and contamination detection—enabling in-depth evaluation and selection of Arabic language models for specific tasks.

Why SILMA AI?

Maximize efficiency and impact

Discover the key advantages of partnering with us

Competency

Our team brings over two decades of multidisciplinary expertise, honed through impactful roles in leading global organizations

Track record

Our track record of delivering robust, effective models speaks for itself — and we’re ready to apply that expertise to your unique challenges

Speed

Our expertise in model training and fine-tuning, backed by experience and reusable assets, helps companies achieve their goals efficiently

Contacts

Contact us

Let us help you harness the power of AI to find innovative solutions. Reach out to us today via hello@silma.ai or via the form

hello@silma.ai