Xiaomi: MiMo-V2.5

Xiaomi's MiMo-V2.5, an omnimodal model created on April 22, 2026, demonstrates competitive response times, ranking in the 40th percentile across eight benchmarks. However, its pricing tends to be at premium levels, placing it in the 19th percentile for cost. A standout feature is its exceptional reliability, achieving a perfect 100% success rate across all benchmarks, indicating minimal technical failures. In terms of performance, MiMo-V2.5 exhibits strong capabilities across various domains. It shows excellent accuracy in Hallucinations (98.0%), Coding (95.0%), General Knowledge (99.5%), and Mathematics (96.0%), with particularly high percentile rankings in Coding (92nd) and Mathematics (92nd). The model achieves perfect 100% accuracy in Email Classification and Ethics, with the latter also being noted as the most accurate model at its price point and among models of similar speed. While its Instruction Following accuracy is solid at 72.0% (77th percentile), its cost for this task is notably low (16th percentile). Reasoning also shows strong performance at 96.0% accuracy. Overall, MiMo-V2.5's key strengths lie in its high accuracy across complex cognitive tasks, particularly in coding, mathematics, and ethical reasoning, coupled with its outstanding reliability. Its primary weakness is its premium pricing structure.