Mixtral 8x22B

Mistral AI's flagship open-weight 141B sparse Mixture-of-Experts model. Apache 2.0 licensed with frontier reasoning, math, and multilingual performance at a fraction of dense compute cost.
Tags
Open Source · LLM · Open Weights · Mistral · Mixtral · MoE · Apache 2.0 · Self-Hostable · Mixtral 8x22B · Free · free tools · free ai tools