April 30, 2025: China Advances AI with Xiaomi and DeepSeek Models - Xiaomi introduced MiMo-7B, open-source reasoning models with 7 billion parameters, which they claim surpass OpenAI's o1-mini in coding and math. The series features a base model and improved versions fine-tuned through supervised learning and reinforcement learning. Meanwhile, DeepSeek upgraded its Prover math AI to Prover-V2, optimized for theorem proving. These advancements coincide with Alibaba's launch of the Qwen3 series, bolstering China's AI landscape with competitive new models.
