France's Mistral announces new AI model
The 1-trillion-parameter model will be tested by developers and security experts before full weights are released later this month, Mistral said.
- On Tuesday, French AI company Mistral launched a public preview of ML4 in Abu Dhabi, with CEO Arthur Mensch claiming the open-weight model outperforms Chinese competitors on cybersecurity.
- Mistral raised €3 billion at a €21 billion valuation in September, the largest equity round completed by a European technology company, positioning itself as a safer alternative to American OpenAI and Anthropic.
- ML4 was trained over two months on 4,000 NVIDIA Grace Blackwell GPUs in Mistral's own European data centers and covers more than 160 languages, including every official EU language.
- Developers and cybersecurity leaders can test ML4 immediately, with full weights releasing October 27. Guillaume Lample, chief scientist and co-founder of Mistral, said the cyber defense capabilities enable enterprises to defend against threat actors jailbreaking closed models.
- The launch highlights the geopolitical edge of open-weight AI, where model parameters can be downloaded and run by anyone, contrasting with closed systems from OpenAI and Anthropic. Mistral's announcement comes as AI agents have been found hacking websites.
104 Articles
104 Articles
Mistral Launches 1-trillion-parameter Large 4 Model, Plans Open-weight Release by Month-end
Get latest articles and stories on Business at LatestLY.model, a 1-trillion-parameter natively multimodal system, with the company set to release its weights by the end of the month as it scales up its European computing infrastructure. Business News | Mistral Launches 1-trillion-parameter Large 4 Model, Plans Open-weight Release by Month-end.
French AI startup Mistral unveiled its latest open-weight AI model, ‘Mistral Large 4 (ML4),’ on the 6th (local time). Mistral emphasized that the performance of the new model is comparable to some Chinese open-weight models. On this day, Mistral stated that ML4 is a large-scale model equipped with one trillion parameters, demonstrating strengths in specialized tasks such as cybersecurity, coding, manufacturing, and finance.
The French company will publish the pesos on October 27. It does not exceed the large closed American models, but aspires to be the best open alternative outside China Read
Mistral's Le Chonk: Europe's 1 Trillion Parameter Bid for AI Independence
Mistral AI launched Mistral Large 4, nicknamed Le Chonk, a 1-trillion-parameter multimodal model with 49 billion active parameters trained in Europe on limited GPUs. It targets cybersecurity and enterprise tasks where some closed models refuse to engage. Weights arrive October 27 after preview testing.
Coverage Details
Bias Distribution
- 49% of the sources are Center
Factuality
To view factuality data please Upgrade to Premium
