WizardLM-2 7B is the smaller variant of Microsoft AI's latest Wizard model. It is the fastest and achieves comparable performance with existing 10x larger opensource leading models. It is a finetune of Mistral 7B Instruct, using the same technique as WizardLM-2 8x22B.