[Xiaomi releases its first large-scale reasoning model MiMo-7B and announces full open source] On April 30, according to Xiaomi's official account "Xiaomi MiMo", Xiaomi today open-sourced its first large-scale model MiMo-7B designed specifically for "Reasoning", surpassing OpenAI o1-mini and Ali Qwen-32B in AIME mathematical reasoning and LiveCodeBench programming assessments. MiMo improves reasoning capabilities through the linkage of large-scale reasoning data pre-training and reinforcement learning post-training, proposes innovative algorithms such as Test Difficulty Driven Reward, and leads in reinforcement learning potential among similar 7B models. The entire series has been open-sourced to Hugging Face, and related technical reports have been released simultaneously.