Mistral claims its open-weight AI model le chonk surpasses chinese competitors

French AI firm Mistral has claimed its open-weight model, Le Chonk, has surpassed leading Chinese models like DeepSeek and Z.ai in cybersecurity and coding capabilities. Mistral released an early version of Le Chonk a few weeks ago, initially positioning it as the most advanced open-weight model outside of China. After further refinement, the company now asserts it is the 'world’s best' for automated coding and cybersecurity. Pierre Stock, VP of science at Mistral, called it a pivotal moment in the global AI leadership race, emphasizing its role in defending against rising cyber threats. Mistral has made Le Chonk freely available, arguing that open-source models are essential for global cybersecurity. Stock warned that Chinese models could be used as threat vectors, citing recent incidents where AI tools uncovered critical security vulnerabilities. He compared the situation to electricity, stating that open-source models are impossible to shut down remotely. However, Rafe Pilling of Sophos noted that open-source models also pose dual-use risks, as they can be exploited by malicious actors regardless of their origin.