Mistral AI demonstrates Mistral Large 4 reverse-engineering malware in 12 minutes

Mistral AI demonstrated Mistral Large 4 reverse-engineering an unknown binary in an end-to-end investigation task. In the demonstration, the model identified the sample as Cobalt Strike, extracted its indicators of compromise (IoCs) and malware configuration, and generated a report including a YARA rule for incident detection.

According to Mistral AI, the model completed the workflow in 12 minutes, compared to an estimated day of manual work.