Data as of Jul 25, 2026 · Based on 2,718,867 AI responses across 9,511 prompts · See how Parse measures this
Magistralmagistral.space · Low evidenceMagistral is a series of AI reasoning models from Mistral AI, trained with scalable reinforcement learning to enhance reasoning, multimodal understanding, and function calling. It includes open-source and medium variants, with published research on arXiv and access via HuggingFace.
Parse Score