Mistral releases Large 4 for cybersecurity — Channel NewsAsia

French startup Mistral AI has opened public access to the Large 4 model, which was developed to help organizations identify and remediate cybersecurity vulnerabilities. Channel NewsAsia reports.

The model can be run independently

Large 4 can be downloaded and run independently, without access through the provider's online service. Mistral said this enables organizations to use the software within their own systems and reduce dependence on third-party providers.

This distribution method distinguishes the model from leading systems such as Google's Gemini, which are mainly available through online services.

More current news is available on the UA.News Telegram channel Telegram.

Update after testing

The company said the released version of Large 4 is an improvement on the model presented three weeks earlier. It was refined after testing with governments and corporate users, which, according to Mistral's assessment, helped improve performance.

Mistral called Large 4 its most powerful openly available model for cybersecurity tasks. The company's vice president of science, Pierre Stock, stressed that after its public release, access to the model cannot be fully restricted: this protects users from a possible shutdown, but at the same time allows anyone to use it.