TMTPOST – iFlytek has announced the full upgrade of its Spark Deep Inference Large Model X1, marking a significant milestone in China’s AI development. The company also introduced the Spark Medical Model X1, built on the newly enhanced X1 architecture. Notably, Spark X1 was trained entirely using domestic computing power and features 70 billion parameters. Despite having significantly fewer parameters than DeepSeek R1 (671B) and OpenAI’s o1, Spark X1 has achieved comparable mathematical reasoning capabilities, even outperforming in certain Chinese-language math tasks. With the X1 upgrade, iFlytek has also updated its AI-driven applications, including Xiaoyi Medical Assistant, Spark Teacher Assistant, and AI Learning Machine, demonstrating advancements in efficiency optimization and industry-specific applications. This breakthrough is expected to accelerate the deployment of deep inference large models in healthcare, education, and other sectors.