Mistral AI has launched a public preview of Mistral Large 4, also known as Le Chonk, featuring 1 trillion parameters and open model weights scheduled for October. The model aims to excel in coding, agentic workflows, and multimodal tasks, with a focus on enterprise applications.
TL;DR
- Mistral AI releases Le Chonk, a 1 trillion-parameter multimodal model, with open weights coming in October.
- NVIDIA and Microsoft launch RTX Spark, a superchip for AI development and gaming on Windows PCs.
- Thoughtworks survey reveals 88% of CIOs believe AI adoption is outpacing governance structures.
What happened
Mistral AI has introduced a public preview of its flagship model, Mistral Large 4, dubbed Le Chonk. This model boasts 1 trillion parameters, with 49 billion active parameters, making it Mistral's largest and most capable model to date. The open model weights are scheduled for release at the end of October.
NVIDIA and Microsoft have announced the RTX Spark, a flagship superchip designed for AI development, creation, and gaming on Windows PCs. The RTX Spark combines an NVIDIA Blackwell RTX GPU with an NVIDIA Grace CPU, offering up to a petaflop of FP4 AI compute. Microsoft's Surface Laptop Ultra is built to support this new superchip.
Why it matters
Le Chonk's advanced capabilities in coding, agentic workflows, and multimodal understanding position it as a strong contender in the enterprise AI market, particularly in domains like cybersecurity, finance, and law. Its open weights will likely foster innovation and customization across key industries.
The RTX Spark represents a significant leap in AI-powered personal computing, enabling users to run complex models on a single Windows PC. This development could democratize AI access and accelerate AI-driven innovation in various fields.
Key facts
- Mistral Large 4 (Le Chonk) has 1 trillion parameters, with 49 billion active parameters.
- Open model weights for Le Chonk will be released at the end of October.
- RTX Spark offers up to a petaflop of FP4 AI compute.
- Microsoft's Surface Laptop Ultra is designed to support the RTX Spark.
- 88% of CIOs believe AI adoption is outpacing governance structures, according to a Thoughtworks survey.
- Reflection AI, backed by NVIDIA, aims to build US open-weight frontier AI models.
- Google Cloud introduces Gemini agent, a universal agent for work that handles knowledge work, creates media, and writes code.
Context
The release of Le Chonk underscores the growing trend of open-weight models, which foster transparency and customization in AI development. This trend is particularly significant in the context of sovereign AI, where local control and adaptability are crucial.
The collaboration between NVIDIA and Microsoft on the RTX Spark highlights the increasing integration of AI capabilities into mainstream computing devices. This development is part of a broader industry shift towards making AI more accessible and user-friendly.
