Skip to content

Translation in progress Some entries still include text in the original language.

Simon Willison· @simonw · X·Indexed by this site · Originally published

Mistral 发布 Mistral Large 4 开放权重模型,1T 参数原生多模态

AI summary

Mistral releases Mistral Large 4, codename Le Chonk, with a natively multimodal architecture of 1T parameters and 49B active parameters. Mistral claims it is the best open-weight model in the US or Europe on aggregate benchmarks, achieves SOTA on critical workloads such as cyber defense, manufacturing, and finance, and surpasses closed-source frontier models in visual grounding. The model was built end-to-end in Europe, can be deployed in Europe via Mistral Cloud, is available via API starting today, and the open-weight version will be released at the end of October, with private collaboration with cybersecurity partners underway. Simon Willison shared the news and demonstrated his pelican test.

Full text · Original

Mistral can pelican now! https://tools.simonwillison.net/markdown-svg-renderer?url=https%3A%2F%2Fgist.github.com%2Fsimonw%2F0db8b8196e57711b352f6c5857cf5b35#response-1

Quoted postMistral AI@MistralAI
Meet Mistral Large 4, aka Le Chonk. • 1T parameters, natively multimodal. 49B active. It is the best open weights model from US or Europe on aggregated benchmarks. • State-of-the-art on critical workloads, including cyber defense, manufacturing and finance and it surpasses closed frontier models on visual grounding. • Forged in Europe end-to-end and is deployable from Europe via our own Mistral Cloud infrastructure. • Available to all via API today. Working with cybersecurity partners privately. Open weights release end of October.
View the quoted post on X

Source: Simon Willison · x.com