Mistral and NVIDIA partner to advance AI for engineering with trillion-parameter model

1. 23:25
   The Decoder

   Mistral releases 'Le Chonk,' a powerful coding model that trails Chinese models overall

   Mistral has released the weights for Mistral Large 4, its largest model to date, nicknamed 'Le Chonk.' The model beats open-source rivals in coding and cyber defense but ranks behind Chinese models overall. After additional training, it improved on the Terminal-Bench benchmark but still trails leading Chinese models like Kimi K3.1 Pro. It offers the best price-performance among open models but ranks below Claude Code in coding capability.

2. 22:00
   TechCrunch · AI

   Mistral releases open-source trillion-parameter AI model ML4 targeting cybersecurity

   Mistral has released the weights for Mistral Large 4 (ML4), a trillion-parameter open source AI model nicknamed "Le Chonk." The release targets enterprises and governments seeking sovereign cyber defense capabilities, offering state-of-the-art performance with 47 percent fewer tokens per task compared to competitor GLM-5.3. The model outperforms other open models on coding agent benchmarks and can be deployed on-premises to ensure data control.

3. 08:00
   Hugging Face BlogSelected

   Mistral releases multimodal MoE model Large 4 (Le Chonk) on Hugging Face

   Mistral Large 4, codenamed Le Chonk, is a multimodal Mixture-of-Experts model with approximately 1 trillion total parameters and 49 billion active parameters. It features a 1-million-token context window and a Pixtral vision encoder, supporting text and image inputs. The model shows strong performance in cybersecurity tasks, scoring 82% on the Artificial Analysis Cyber Index test, and is positioned for defensive security work and coding agents. It is released under the Mistral Ecosystem Open-Weight License and available in FP8, NVFP4, and GGUF formats.

1. 22:40
   The Decoder

   Mistral and Nvidia team up to speed up industrial design with physics AI

   Mistral and Nvidia are combining Mistral Large 4, a one-trillion-parameter language model trained on 3,800 Blackwell GPUs, with Nvidia's CUDA-X, Warp and Omniverse libraries to accelerate engineering simulations and industrial product design. The partners aim to link design, simulation, and AI training into a continuous workflow for sectors including aerospace, automotive, and semiconductors, and plan to release public benchmarks and open datasets for testing physics models.

2. 20:00
   Mistral AISelected

   Mistral and NVIDIA partner to accelerate engineering AI with open benchmarks

   Mistral and NVIDIA are collaborating to integrate Mistral's AI models with NVIDIA's accelerated computing and physics tools, enabling faster simulations and closed-loop AI-driven design. The partnership will introduce public benchmarks and open datasets to standardize evaluations for physics and agentic AI in computer-aided engineering.

The chart compares consistently observed participants only. Hover or tap to inspect each hour; use arrow keys to move between observations.