Mistral Releases Le Chonk as Top Open AI Model for Cybersecurity — SkimNews
Get the Tech newsletter
Daily tech — startups, AI labs, chips, the launches that shape the next decade. Free.
- Mistral released the finalized version of its AI model Le Chonk as a free, open-weight model, claiming it is now the world’s best for automated coding and cybersecurity tasks
- Mistral says Le Chonk, a trillion-parameter model, matches or exceeds performance from leading Chinese labs including DeepSeek and Z.ai after training on progressively harder coding and security challenges
- Pierre Stock, VP of science at Mistral, argues open-weight models are essential for Western cyber defense because they cannot be remotely disabled, unlike proprietary models from OpenAI, Anthropic, or Google
- Rafe Pilling of Sophos acknowledges open-weight models like Le Chonk are dual-use, noting freely available models can be exploited by threat actors despite their defensive potential
- Mistral positions Le Chonk as a response to rising AI-driven cyber threats, citing incidents where AI models discovered critical flaws in foundational web technology and even breached corporate systems
- Hacker News and TechMeme confirm the release of Le Chonk’s weights, with TechMeme highlighting its competitive standing against top open models like GLM-5.3 on coding benchmarks
Why it matters: Western governments and enterprises gain access to a high-performance, sovereign AI model for cybersecurity at a time of rising concern over foreign-controlled AI, but the open release also lowers the barrier for malicious actors to weaponize advanced capabilities, creating a strategic trade-off between resilience and risk.
Ask SkimNews
