AI is definitely the trend this year, and Xiaomi is the latest player to enter the game. Xiaomi has unveiled MiMo, its first in-house large language model (LLM) and here's what we know about it so far.
MiMo was created by Xiaomi’s Core AI team using reinforcement learning and is an open-source LLM with 7 billion parameters. Xiaomi reportedly acquired 10,000 GPUs to train its AI models, and MiMo can outperform OpenAI’s o1-mini and Alibaba’s QwQ-32B-Preview in math reasoning and coding.
In addition, it was also reported that Xiaomi attempted to recruit AI prodigy Luo Fuli from DeepSeek, though she declined. And what will Xiaomi use MiMo for? Well, the report said the company plans to embed its AI into EVs, smartphones, and appliances, aiming long-term for artificial general intelligence (AGI).
To the uninitiated, AGI is a theoretical point at which AI equals or surpasses human intelligence, and Xiaomi thinks the road to AGI is still very long. The year 2025, particularly the second half of the year, is all about AI model competition, which we can expect to see a full-out battle between big tech firms.
COMMENTS