Mistral AI launches Large 4 model for cybersecurity users

French AI startup Mistral AI has released its Large 4 model publicly, TechGig reported. The model is designed to help organisations identify and fix cybersecurity vulnerabilities. As FTMQ SI reported earlier, the firm previously made the open weights for the model available to enterprise software developers and cybersecurity users. [1][10]

TechGig reported that the public release is an upgrade from a version unveiled three weeks prior following testing with governments and corporate users. Mistral AI stated that this testing helped improve the performance of the model. Pierre Stock, Vice President of Science at Mistral, described Large 4 as the most capable openly available model from the company for cybersecurity use. [1]

Unlike competing AI systems such as Google Gemini, the Mistral model can be downloaded and run independently, TechGig reported. Mistral stated that running the model independently reduces dependence on third-party providers. [1]

Mistral AI was founded in 2023 and is headquartered in Paris, France. As FTMQ SI reported earlier, the Large 4 model features a mixture of experts architecture, a 1 million token context window, and 1.05 trillion total parameters. [7][10][11]