Mistral 7B & Mixtral
AI Model
Language
Historical
The small models that punched up — and mainstreamed open MoE.
You're looking at an AI model record: who built it, what it can do, and the hardware and data behind it.
Mistral 7B (September 2023) beat much larger models at a fraction of the cost; Mixtral 8x7B (December 2023) was the first widely used open mixture-of-experts, prefiguring the architecture DeepSeek and Qwen scaled up. Together they made 'small but smart' a strategy and put Europe on the map.
How this connects to the rest of the stack
This record's links at a glance — click a group to jump to it.
Developed by
Based on
Succeeded by
Sources
Record reviewed as of 2026-01-15 — see Methodology.