Mistral 3: unveiling the new generation of open-weight models

Mistral AI is introducing the Mistral 3 model family, which includes the Large version with MoE architecture and the Ministral series for local devices. All models are available under the Apache 2.0 license.

Loading the Elevenlabs Text to Speech AudioNative Player…

Models with open weights are becoming increasingly competitive with closed commercial solutions. Mistral AI just unveiled the third generation of its products — the Mistral 3 family. Its launch is a step toward realizing the vision of distributed intelligence.

What does this mean in practice? Until now, to use the highest-quality AI, we typically had to connect to a powerful cloud server. The vision of distributed intelligence changes this approach. The idea is for AI to run where it’s needed, not only in a distant data center. With the different sizes of Mistral 3 models, developers can choose the right tool for the task: a gigantic model for complex analyses in the cloud, or a small, fast model that runs directly on your laptop, even without internet access.

Ten artykuł jest częścią płatnej edycji hAI Magazine. Aby go przeczytać w całości, wykup dostęp on-line

25 zł miesięcznie

Wykup dostęp

Współtwórczyni newslettera AI Flash, studentka psychologii i pasjonatka sztucznej inteligencji. Interesuję się wpływem nowych technologii na człowieka, a w wolnych chwilach eksperymentuję z generatywną grafiką w Midjourney.

Share

You might be interested in