36 Krypton learned that recently, Zhongke Shuguang released a new iterative upgrade of AI inference native storage FN Neo. The measured data shows that the response time of the first token of the model is reduced by 73%, and the overall token throughput capacity is increased by 150%, which effectively reduces the waiting time for user interaction and strengthens the concurrent service capability of the system. FN Neo also has 99.999% high reliability for the uninterrupted operation of 7 × 24 hours of production-grade AI business, and the multiple failure prevention mechanism ensures uninterrupted inference business continuity. At the same time, the product is compatible with the mainstream reasoning framework and the multivariate networking environment. Enterprises can complete the deployment without large-scale transformation of the existing system, improve quality and efficiency under the premise of reusing the existing reasoning ecology, and effectively support the large-scale production of large models and agents.
Hangtong
Hangtong Technology