Top Turkish AI Models: 10 Artificial Intelligence Models Developed in Turkey

Turkey is becoming an important contributor to artificial intelligence development. Local technology companies, universities, research institutions, and government organisations are investing in AI technologies. Their work includes large language models (LLMs), conversational AI, natural language processing, machine learning, and computer vision.

Many Turkish AI models focus on understanding and generating the Turkish language. They are designed to handle Turkish grammar, sentence structures, expressions, and cultural context. Some also support English and other languages.

These models serve different purposes. Some are designed for general conversations and content generation, while others focus on research, business automation, translation, and specialised applications.

Below are 10 notable Turkish AI models and projects, their developers, features, and main strengths.

Comparison of Turkish AI Models

The following table highlights some of the notable artificial intelligence models developed by Turkish companies, universities, and research organisations.

1. Kumru AI

Kumru AI is a Turkish artificial intelligence model developed by VNGRS. It focuses on understanding and generating Turkish text. The project aims to address the language-specific challenges that general-purpose international AI models may experience when processing Turkish.

Kumru includes different model sizes, including the Kumru-2B model. The smaller version is available with open weights, while other versions have separate access arrangements.

Features:

• Turkish-focused language processing.

• Text generation and conversational capabilities.

• Support for summarisation and language understanding.

• Smaller model versions for local deployment.

• Open-weight availability for selected versions.

Strengths:

Kumru's main strength is its focus on Turkish. It is designed to understand Turkish expressions and sentence structures. Its smaller model versions also make it useful for developers interested in running AI systems with fewer computing resources.

2. Trendyol LLM

Trendyol LLM is an artificial intelligence model family developed by Trendyol, one of Turkey's major e-commerce technology companies.

The models are designed for language generation, conversational applications, and business-related tasks. The family includes several versions with different architectures and capabilities.

Trendyol has also developed specialised AI models for areas such as cybersecurity, image understanding, and e-commerce.

Features:

• Turkish and English language processing.

• Text generation and conversational AI.

• Multiple model sizes and versions.

• Reasoning capabilities in selected releases.

• Specialised applications for e-commerce.

• Open-weight releases for developers.

Strengths:

Trendyol LLM benefits from its developer's experience in handling large-scale online business operations. Its models are useful for customer support, product descriptions, business automation, and other commercial applications.

The wider Trendyol AI ecosystem also provides specialised models for image and product-related tasks.

3. Turkcell LLM

Turkcell LLM is an artificial intelligence project developed by Turkcell, a major Turkish telecommunications company.

The model focuses on Turkish language understanding and text generation. It is part of the company's wider investment in artificial intelligence, digital services, and language technologies.

One notable release is Turkcell-LLM-7b-v1, which is based on the Mistral architecture.

Features:

• Turkish text generation.

• Natural language understanding.

• Conversational AI applications.

• Support for AI research and development.

• Open-weight availability for its published model.

Strengths:

Turkcell LLM is useful for developers exploring Turkish-language applications. Its telecommunications background also makes the project relevant to customer communication, digital assistants, and automated service systems.

Its published model provides an opportunity for researchers to experiment with Turkish-focused language applications.

4. Turkish-Gemma

Turkish-Gemma is a family of Turkish language models developed by researchers associated with Yildiz Technical University's COSMOS research group.

The models use Google's Gemma architecture and are adapted for Turkish language applications.

One notable release is Turkish-Gemma-9b-v0.1.

Features:

• Turkish-focused language processing.

• Text generation and comprehension.

• Research-oriented model development.

• Adaptation of the Gemma architecture.

• Support for experimentation and fine-tuning.

Strengths:

Turkish-Gemma benefits from the underlying Gemma model architecture while receiving additional Turkish-language training.

It is useful for researchers and developers who want to build applications that require better Turkish language understanding.

5. TURNA

TURNA is a natural language processing project associated with Boğaziçi University and its TABILab research group.

The model is designed for language-related research and text processing. It is particularly relevant to tasks such as translation, summarisation, and language generation.

Features:

• Turkish language processing.

• Text summarisation.

• Machine translation applications.

• Research-oriented development.

• Encoder-decoder architecture.

Strengths:

TURNA is valuable for academic research and experimentation. Its focus on language processing makes it suitable for developers working on translation systems, text analysis, and other natural language processing applications.

Its academic licensing conditions should be checked before commercial use.

6. Kanarya

Kanarya is a Turkish language model associated with Koç University's research ecosystem.

The project contributes to the development of language technologies that can understand and generate Turkish text. Its Kanarya-2B release is an example of a smaller language model developed for Turkish-focused research.

Features:

• Turkish text generation.

• Language modelling.

• Research and experimentation.

• Smaller model architecture.

• Support for Turkish NLP development.

Strengths:

Kanarya provides researchers with another option for studying Turkish language models. Its relatively small size also makes it interesting for experimentation in environments with limited computing resources.

7. T3 AI

T3 AI is an artificial intelligence initiative associated with the Turkish Technology Team Foundation (T3 Foundation) and its technology partners.

The project aims to support the development and use of artificial intelligence technologies in Turkey.

Its ecosystem focuses on conversational AI and applications that can serve Turkish users.

Features:

• Turkish conversational AI development.

• Language-based applications.

• AI research and innovation.

• Community participation.

• Integration with technology development initiatives.

Strengths:

T3 AI benefits from its connection to Turkey's technology and innovation ecosystem. It supports local AI development and encourages participation from researchers, developers, and technology enthusiasts.

Its availability and individual model specifications should be checked against current official project announcements.

8. BİLGE

BİLGE is a national artificial intelligence development project associated with TÜBİTAK BİLGEM, a Turkish research and technology institution.

The project focuses on developing domestic AI capabilities and language technologies. It is part of Turkey's wider efforts to strengthen its technological independence.

Features:

• Turkish language AI development.

• Large language model research.

• National technology development.

• Potential applications in public institutions.

• Research into advanced AI capabilities.

Strengths:

BİLGE's main strength is its institutional research backing. It represents an investment in developing AI technologies that can serve Turkish organisations and national technology projects.

General public access and deployment options depend on the availability of individual releases.

9. KoçDigital LLM

KoçDigital LLM is an AI language model project associated with KoçDigital.

The project contributes to the development of language technologies and enterprise AI applications.

It is particularly relevant to organisations interested in using artificial intelligence to improve business processes.

Features:

• Language model development.

• Turkish language applications.

• Enterprise AI research.

• Business automation opportunities.

• Support for specialised AI applications.

Strengths:

Its main strength is its connection to enterprise technology. Business organisations can benefit from language models through automated communication, document processing, and information management.

Developers should check the availability and licensing of individual releases before selecting the model.

10. WiroAI Turkish LLM

WiroAI Turkish LLM is a Turkish language model developed by WiroAI.

The project focuses on Turkish text generation and language-based AI applications. Its published model family includes Turkish language models and reasoning-oriented releases.

Features:

• Turkish language generation.

• Conversational AI applications.

• Language understanding.

• Reasoning-focused model development.

• Open model releases.

Strengths:

WiroAI provides developers with additional options for building Turkish-language AI applications. Its reasoning-oriented models are particularly interesting for people researching AI systems that can handle more complex questions and problem-solving tasks.

How to Choose a Turkish AI Model

Choosing the right Turkish AI model depends on the intended application.

Consider the following factors:

It is also important to check model size, hardware requirements, licensing, API availability, and commercial usage permissions.

Some models are available for direct experimentation, while others are research projects or platforms with restricted access.

Frequently Asked Questions

1. Which Turkish AI model is best?

There is no single best model for every task. Kumru, Trendyol LLM, and Turkish-Gemma are interesting options for Turkish language applications.

2. Are Turkish AI models available for free?

Some Turkish AI models have open weights that developers can download and use under specific licences. Other models require registration or commercial access.

3. Can Turkish AI models understand English?

Several models support both Turkish and English. However, their language capabilities differ depending on their training data and architecture.

4. Can businesses use Turkish AI models?

Yes. Businesses can explore these models for customer support, document processing, content generation, translation, and workflow automation. Licensing conditions must be checked first.

5. Are Turkish AI models open source?

Some have publicly available weights and source code, while others are proprietary or have restricted access. Open weights do not always mean that every part of a model is open source.

The Growing Turkish AI Ecosystem

Turkey's artificial intelligence industry includes contributions from technology companies, universities, government research institutions, and independent developers.

Projects such as Kumru, Trendyol LLM, Turkcell LLM, and Turkish-Gemma demonstrate the growing interest in Turkish-focused language technology. Academic projects such as TURNA and Kanarya also support research and experimentation.

These developments provide more choices for businesses, researchers, and developers interested in Turkish language AI.

As investment and research continue, Turkish AI models are expected to play a growing role in language processing, business automation, and specialised digital services.

› More Article