🧠 Introduction to Kumru.ai

Kumru.ai is Turkey’s first fully locally trained large language model (LLM), developed by the technology company VNGRS. Designed to serve Turkish-speaking users and institutions, Kumru is part of Turkey’s national AI strategy, aiming to reduce dependence on foreign models and enable secure, scalable, and high-performance AI deployments across sectors.

🔑 Key Features

• Locally Trained LLM: Built from scratch using 500GB of curated Turkish-language data.
• Model Architecture: 7.4 billion parameters, decoder-only architecture similar to Mistral and LLaMA-3.
• Context Window: Supports up to 8192 Turkish tokens—ideal for long-form documents and enterprise use.
• Multilingual Support: Optimized for Turkish, but also performs well in English and programming tasks.
• Lightweight Deployment: Runs on 16GB GPUs, enabling on-premise deployment for banks, hospitals, and government agencies.
• Enterprise-Ready: Suitable for document processing, question answering, customer support, legal analysis, and more.

🎯 Ideal Use Cases

• Public Sector & Government: Ensures data residency and sovereignty with local deployment.
• Finance & Healthcare: Enables secure document analysis, compliance workflows, and AI-powered support.
• Education & Research: Supports academic research, Turkish-language processing, and AI literacy.
• Developers & Integrators: Can be embedded into applications for localized AI capabilities.

🌍 Why Kumru Matters

Kumru represents a strategic leap in AI independence for Turkey. It empowers institutions to deploy AI systems that understand Turkish deeply, operate securely within national infrastructure, and scale across industries. With its open architecture and lightweight footprint, Kumru is a practical and powerful alternative to foreign LLMs.