Cisco Antares SLMs: Smaller AI for Enterprise Security

Cisco has launched Antares, a new family of small language models, or SLMs. These models are designed to help enterprise software teams find security flaws. Here's the thing: using large AI models for security can be expensive due to token costs. There's also the risk of sending private source code to cloud services. Antares aims to solve these problems. Cisco is releasing Antares-350M and Antares-1B as open-weight models. They are built to pinpoint vulnerable code in large software, using fewer tokens than general-purpose models. What's more, they can run locally, keeping sensitive code within customer control. A Cisco AI Researcher noted that while powerful, frontier models aren't always the best fit for specialized security tasks. Antares came from research showing that smaller models can effectively learn search strategies. The bottom line: this release shows a shift towards domain-specific AI models that offer targeted capabilities, lower costs, and better data governance for enterprises. This could significantly reduce investigation time for security teams.