February 10th.Silicon-based flowAnnounced online today DeepSeek R1 & V3 enterprise-level services provide four enterprise-level service models for customers to choose from: Dedicated Instance, Arithmetic NanoManager, Private MaaS, and Huawei Rise 910B-based All-in-One PC.
1AI learned from the official presentation that"Dedicated instance" service modelMainly for enterprise-level application scenarios. A single instance can be deployed with a full-blooded version of the DeepSeek R1 or DeepSeek V3 model.Exclusive use of Huawei's Rise 16 card computing power, with a service fee of 90,000 RMB/month.. To ensure high availability and seamless failover of the system, Silicon Mobility recommends procuring at least 2 instances to ensure stability and continuity of service.
In addition to the full-blooded DeepSeek R1 & V3 models, Silicon Mobility also provides proprietary instances of the DeepSeek-R1 distillation model, Qwen, CozyVoice, FLUX.1, and other types of models already available on the SIliconCloud platform.
"Arithmetic in Care" service modelIt is mainly for enterprises with a certain scale of their own computing power (which needs to be connected to the public network) that want to run a MaaS service (SiliconCloud) for silicon based mobility on their own computing power. The service helps customers build customer-specific clusters of SiliconCloud and provide services to customers with that computing power.
"Private MaaS" service modelIt is mainly targeted at enterprises with a certain scale of their own arithmetic that want to deploy Silicon Flow Private Edition MaaS in a private and isolated arithmetic environment, and we support the deployment of models and inference services on the SiliconCloud platform.
All-in-One PC based on Huawei Rise 910BPrimarily targeted at enterprise customers who want to keep their data local, the all-in-one product comes with a built-in inference acceleration framework and model plaza (including, but not limited to, the full range of DeepSeek models) from Silicon Flow Online.