Mistral Launches Public Preview of ML4, Its Largest AI Model Yet

Oct 6, 2026

AI Summary

Mistral has introduced a public preview of its latest AI model, ML4, which features 1 trillion parameters and excels in various applications, particularly in cybersecurity and coding. The model is set to be officially released by the end of the month, with capabilities that surpass many existing models in both open and closed categories.

Mistral has launched a public preview of ML4, its largest AI model with 1 trillion parameters and 49 billion active parameters.
ML4 demonstrates strong performance in coding, cybersecurity, and multimodal understanding, outperforming many existing open-source models.
The model was trained on 3,800 NVIDIA GPUs in Mistral's European data centers and will be available globally, including a European deployment compliant with local laws.
ML4 ranks among the top five models on the Artificial Analysis Cyber Index for its ability to find and fix security flaws, achieving an 82% success rate in reproducing and patching vulnerabilities.
It excels in software engineering tasks, scoring well on various benchmarks, and is particularly strong in visual grounding and scientific capabilities.
The model's release is part of Mistral's broader investment strategy, supported by a €3 billion Series D funding round, aimed at expanding compute capacity and enhancing AI capabilities.
Further details on ML4's architecture and performance benchmarks will be shared upon the official release of the model weights at the end of the month.

mistrallanguage modelsai researchnlpopen source

Mistral Launches Public Preview of ML4, Its Largest AI Model Yet

Related Stories

Launch of Claude Haiku 5.5: A Cost-Effective and Fast AI Model

OpenAI Introduces GPT-6 and User-Friendly Interface

Mistral AI launches new large multimodal model to compete with global AI rivals

Mistral Launches ML4, a 1 Trillion-Parameter AI Model with Advanced Capabilities