French AI company Mistral has released the weights for Mistral Large 4 (ML4), a trillion-parameter model that organizations can download and run within their own systems. The move gives businesses and public institutions more control over how they deploy advanced AI.
Mistral says ML4 delivers performance comparable to the open-weight model GLM-5.3 while using 47% fewer tokens per task, which could make some workloads more cost-efficient. The company is positioning the model for cybersecurity, including helping teams analyze code and IT systems and coordinate responses.
Because the weights are available for organizations to operate themselves, deployment can take place within their own environments. Mistral says this could support greater autonomy in applying AI to security work, while enabling institutions to tailor the system to their needs.
Making advanced AI more widely deployable could help organizations build security capabilities that are more adaptable and locally controlled.
