Deploy DeepSeek Models on Alibaba Cloud PAI with One-Click: A Step-by-Step Guide

Step 1: Open Model Gallery

Log in to the PAI console, select the appropriate region, open the workspace, then choose Quick Start → Model Gallery .

Step 2: Choose a DeepSeek model

In the Model Gallery list, click the desired model card, e.g., “DeepSeek‑R1‑Distill‑Qwen‑7B”, to open its detail page.

Step 3: One‑click deployment

Click the Deploy button at the top right. DeepSeek‑R1 can be accelerated with vLLM; DeepSeek‑V3 supports vLLM and Web‑app deployment; the distilled DeepSeek‑R1 model can use BladeLLM or vLLM. After selecting the deployment method and resources, a PAI‑EAS service is created.

After deployment succeeds, view the endpoint and token on the service page, and refer to the model’s documentation for usage instructions.

For cost‑effective integration, this tutorial uses the distilled DeepSeek‑R1‑Distill‑Qwen‑7B model, which transfers DeepSeek‑R1 inference capability to a smaller Qwen model, reducing compute cost while maintaining performance. The Model Gallery also offers one‑click deployment of the original DeepSeek‑R1 and DeepSeek‑V3 models.