iFlytek claims its latest LLM rivals OpenAI and DeepSeek

Chinese artificial intelligence pioneer iFlytek announced a major upgrade to its proprietary deep reasoning model, Spark X1, the industry's only large language model trained entirely on China's domestic computational infrastructure.

The enhanced Spark X1 demonstrates significant improvements across general AI tasks, including mathematics, coding, logical reasoning, text generation, language understanding and knowledge-based Q&A, iFlytek said.

Remarkably, despite being smaller in parameter size compared to leading global counterparts, Spark X1 now rivals the performance of OpenAI's o1 and DeepSeek's R1 models, iFlytek added.

By incorporating diverse, scenario-specific training data, Spark X1 has an edge in key sectors such as education, healthcare and legal services, delivering industry-leading solutions.

Amid intensifying global tech competition, Spark X1 is fully trained on AI chips developed by Huawei Technologies Co, iFlytek and Huawei's AI chip joint research team have been working closely to come up with better domestic computing solutions for AI training, iFlytek added.