The next generation of AI-specific Vera CPUs are said to be twice as fast as other CPUs

On 22 July, Ying Weida wrote that DeepInfra ' s baseline tests showed that NVIDIA Vera CPU was more than twice as fast as other CPUs and could support more AI agents. DeepInfra, the cloud platform, a participant in the early experience of the OpenAI ecosystem of NVIDIA, independently designed and operated baseline tests based on its production-based AI proxy infrastructure. DeepInfra processes nearly 5 trillion coins a week, about 30 per cent of which are driven by the agency system. Its cloud platform is dedicated to high-throughput AI reasoning. The baseline test showed that, with the same quality of service, co-agents supported up to 1.6 times, with a coordinated rate of 2.2 times, while increasing infrastructure utilization and cost-effectiveness. These results show that NVIDIA Vera CPU can achieve the cost-effectiveness, low delay and throughput required by the production agent ' s intelligence. As AI agents assume more complex reasoning, planning, use of tools and data flows, CPUs become increasingly important in coordinating efforts around each model. Vera CPU was designed for proxy work loads as part of the British Weedda extreme joint design strategy for the AI plant. DeepInfra ' s benchmark test highlighted how NVIDIA Vera CPU helped cloud service providers to improve infrastructure utilization, improve cost efficiency and support more co-dispatch AI agents with the same quality of service。