Cisco Challenges GPT-5.5 with Its Open Source AI Models in Cybersecurity
Le brief IA que les pros lisent chaque soir
Les 7 actus IA du jour, décryptées en 5 min. Gratuit.
Inclus dès l'inscription : notre sélection des meilleurs guides & comparatifs IA.
Choisis ton rythme
Gratuit · Pas de spam · Désabonnement en 1 clic
Cisco Challenges GPT-5.5 with Its Open Source AI Models in Cybersecurity
Cisco has launched two small open source AI models for cybersecurity: Antares-350M and Antares-1B, which detect vulnerabilities in software code. Cisco's argument is based on cost-effectiveness. Developer Aman Priyanshu claims on X that the smaller model detects about 150 times more vulnerabilities per dollar than larger AI agents like Cognition's Devin Security Swarm. According to Cisco's internal tests, Antares scanned 500 code repositories in about 15 minutes for less than one dollar, while GPT-5.5 required five hours and cost over 100 dollars for the same task.
The Antares models operate locally, ensuring that sensitive code never leaves the company. The technical report indicates that the models were trained on approximately 72% of data related to security concepts and 15% of code search histories.
Cisco retains a larger version with three billion parameters for its own products. This version is said to have performance close to that of GPT-5.5 and would outperform open source models by up to 200 times its size. The company is also exploring the creation of an industrial consortium for open source AI security tools.
Brief IA — L'actualité IA en français
L'essentiel de l'actualité de l'intelligence artificielle, décrypté et expliqué chaque jour.