Deploying DeepSeek-R1 Models on Tencent Cloud HAI Platform

This article introduces how to deploy DeepSeek-R1 large language models on Tencent Cloud's HAI platform. The platform allows developers to access DeepSeek-R1 models in just 3 minutes without the need for complex setup procedures like purchasing GPUs, installing drivers, configuring networks, or setting up environments.

The deployment process involves two simple steps: First, log in to Tencent Cloud HAI and create a DeepSeek-R1 application (first-time users need to complete authorization). Second, after creation, developers can access the model through either a visual interface (ChatbotUI) or command line (JupyterLab) using credentials obtained via internal messages.

For visual interface access, users navigate to the HAI console, select 'Compute Connection' → 'ChatbotUI', and follow the on-screen instructions. For command line access, users select 'Compute Connection' → 'JupyterLab', open a new Terminal, and run the command 'ollama run deepseek-r1' to load the default 1.5B model. Users can also specify different model sizes (7B/8B/14B) by adding parameters to the command.

The HAI platform also integrates with other Tencent Cloud services like Cloud Studio and Object Storage, enabling developers to quickly build enterprise-grade AI applications. Additionally, Tencent Cloud TI supports R1 and V3 model deployments.