AI Development
Release of Mixtral 8x7B (Mixture-of-Experts)
Editorial Analysis
Mistral popularized the Sparse Mixture-of-Experts (SMoE) architecture with Mixtral 8x7B. It achieved GPT-3.5 levels of performance while being open-weight, setting a new standard for efficient scaling in the open-source community.
At a Glance
Date
December 11, 2023
Importance
High
5/5
Category
model
Axis of Change
Capability Gain
Organizations
Models Affected
Sources