TokenAI to Release Horus Cyber Nano AI Model
Egypt-based company says its new AI model can help security teams find software weaknesses, review code and support simulated cyberattacks.
TokenAI plans to release Horus Cyber Nano 1.0 on September 3, 2026. The company designed the model specifically for cybersecurity tasks, including secure code review, vulnerability analysis and red-team testing.
Unlike a general-purpose AI chatbot, Horus Cyber Nano focuses on work that security and IT teams perform when they test or protect digital systems. It can review software code for weaknesses, analyze technical information and help teams investigate possible attack paths.
What Horus Cyber Nano can help with
-Finding weaknesses in software code
-Reviewing potential security risks
-Supporting penetration testing and red-team exercises
-Analyzing technical reports and system information
-Helping security teams investigate possible attack paths
Smaller Model Targets High-End Performance
TokenAI says Horus Cyber Nano already performs close to much larger AI systems in several public comparisons. Some of those competing systems are roughly 100 times larger, according to the company.
This could matter for organizations that want powerful AI tools without the infrastructure requirements of the largest models. Smaller models can require less computing capacity and may offer more practical deployment options.
TokenAI has published early results from several widely used AI benchmarks. These tests cover areas such as reasoning, software engineering and terminal-based tasks. However, they do not measure cybersecurity performance directly, those results are yet to be published.
The company will make Horus Cyber Nano available as an open-weights model. This means organizations will be able to download and run the model on their own infrastructure, subject to TokenAI’s license terms. For security-sensitive users, local deployment could reduce the need to send code or technical data to an external AI service.
