CloudSky encourages flying: strive to reduce the reasoning cost of one million tokens by more than 100 times.

On February 3, Yunjian Lifly officially held the "Big Computing Chip Strategic Outlook Conference", publicly disclosing the strategic layout of large-scale AI inference chips for the next three years for the first time. The company's executives announced that in the future, Yunjian Lifly will take the GPNPU architecture as the core, vigorously promote the strengthening of cloud-based large computing power through software and hardware collaboration and storage system breakthroughs, strive to reduce the cost of inference for millions of Tokens by more than 100 times, and promote large models from demonstration applications to large-scale delivery. It was introduced that the company underwent a restructuring last year, forming four major business units, among which the government and enterprise business unit will strategically focus on the construction of inference equipment and intelligent computing centers. In addition, Yunjian Lifly also plans to create regional-level "Thousand-Calorie Clusters" to set a new benchmark for AI computing power in cities.