Mistral Claims Its 'Le Chonk' Model Outperforms Top Chinese Open

Mistral Claims Its 'Le Chonk' Model Outperforms Top Chinese Open-Weight Rivals


The French AI lab says its latest open-weight release matches or beats leading models from DeepSeek and Z.ai in coding and cybersecurity benchmarks.


By Joel Khalili | October 29, 2026




A New Contender in the Open-Weight Arena


Mistral, the Paris-based AI lab, has unveiled a new open-weight model it calls Le Chonk, claiming it surpasses the best openly available models from China—including those from DeepSeek and Z.ai—on key benchmarks for coding and cybersecurity.


The announcement, made on October 29, 2026, positions Mistral as a leading Western challenger in the rapidly intensifying open-weight race, where Chinese labs have held considerable momentum over the past two years.


Benchmarks and Claims


According to Mistral, Le Chonk matches or outperforms DeepSeek's and Z.ai's flagship open models on a range of code-generation and security-focused evaluations. The company has not yet released full technical details or independent verification, but the claim signals growing confidence from European AI developers.


"We believe open-weight models should not be defined by a single region," a Mistral spokesperson said. "Le Chonk demonstrates that European innovation can compete at the highest level."


The 2026 Open-Weight Landscape


Le Chonk arrives at a pivotal moment. In 2026, open-weight models have become the backbone of enterprise AI adoption, favored for their transparency, cost-efficiency, and on-premise deployability. Chinese labs like DeepSeek and Z.ai have dominated community leaderboards for coding and security tasks, often releasing models under permissive licenses.


Mistral's new release aims to shift that balance, offering a European alternative that it says delivers comparable—or superior—performance without sacrificing openness.


What's Next


Mistral plans to publish Le Chonk's model card and evaluation results in the coming weeks. The AI community will be watching closely to see whether the French lab's claims hold up under independent testing—and whether Le Chonk can genuinely leapfrog the Chinese open-weight ecosystem.


This is a developing story. WIRED will update as more information becomes available.

via Wired AI

Related