Mistral AI releases cybersecurity-focused Large 4 model to the public

French AI startup Mistral AI publicly released its Large 4 model on Thursday. The model is designed to help organizations identify and remediate cybersecurity vulnerabilities. Unlike leading AI systems accessed via online services such as Google's Gemini, this model can be downloaded and run independently by anyone.

Mistral AI stated that the released version is an upgrade over the version introduced three weeks ago. During this period, tests conducted with governments and corporate users contributed to improving its performance. The company noted that organizations can use the software within their own systems, thereby reducing reliance on a third-party provider.

Competing in a field dominated by Chinese AI models, Mistral described this release as the most capable open-access model to date for cybersecurity use. Pierre Stock, Mistral's Vice President of Science, stated that the model's open weights ensure that access cannot be restricted and prevent users from being excluded.