Chinese tech giant Xiaomi has quietly launched a trillion-parameter artificial intelligence model, the MiMo v2 Pro, which performed so impressively in initial tests that it was mistakenly identified as the upcoming DeepSeek V4, according to industry analysts. The release marks Xiaomi’s significant entry into the large language model arena, challenging established players like OpenAI and Google, and underscoring China’s growing prowess in AI technology.
Xiaomi, best known for smartphones and consumer electronics, has been investing heavily in AI research over the past few years. The MiMo v2 family, including the Pro version, represents a leap in scale, with parameters rivaling those of the largest models globally. “We observed capabilities that matched top-tier models, leading to initial identification errors,” said an analyst from a major tech research firm, who spoke on condition of anonymity. “Xiaomi has clearly made significant strides in AI development.”
The confusion with DeepSeek V4 highlights the rapid pace of innovation in China’s AI sector. DeepSeek, backed by investments from tech giants, is expected to release its V4 model soon, and the similarity in performance metrics has sparked discussions about benchmarking standards and transparency. Experts note that reaching a trillion parameters is a key milestone, often correlating with improved performance on complex tasks.
Looking ahead, Xiaomi’s entry could intensify the global AI race, prompting increased investment and innovation. However, it also raises questions about resource allocation, as trillion-parameter models require substantial computational power and energy, contributing to environmental concerns. Western companies may face heightened competition, potentially accelerating efforts to develop more efficient and ethical AI systems.