Cisco Systems Inc. released Antares, a family of small, open-weight language models designed to scan source code for known security vulnerabilities on Tuesday [1].
This release addresses the growing need for affordable security tools as companies struggle to manage the scale of modern software development. By providing open-weight models, Cisco allows organizations to implement vulnerability scanning without the high costs associated with massive, proprietary AI systems [1, 2].
The Cisco Foundation AI developed the Antares models to help developers locate flaws within their codebases more efficiently [1]. The company made two models available as open-weight downloads [2]. These models are hosted on the Hugging Face platform, allowing for broader accessibility, and community integration [2].
Small language models are specifically designed to be more computationally efficient than their larger counterparts. This efficiency reduces the hardware requirements and energy costs for companies running the scans internally [1]. The primary goal of the Antares project is to provide a low-cost solution for improving overall source-code security [1, 2].
By focusing on known vulnerabilities, the AI helps security teams prioritize patches and identify risky patterns before they can be exploited by attackers. The open-weight nature of the release means that developers can potentially fine-tune the models for specific programming languages, or internal coding standards [2].
Cisco said the models are intended to lower the barrier to entry for robust security auditing [1]. The availability of these tools on Hugging Face enables a faster deployment cycle for security teams who can now integrate the models directly into their existing continuous integration and continuous deployment pipelines [2].
“Cisco released Antares, a family of small, open-weight language models designed to scan source code for known security vulnerabilities.”
The shift toward small, open-weight models indicates a broader trend in AI where specialized, efficient tools are replacing general-purpose giants for enterprise tasks. By focusing specifically on vulnerability detection, Cisco is attempting to democratize security auditing, moving it from a high-cost luxury to a standard part of the development lifecycle.
