Deploy OpenAI’s gpt-oss-20b on Alibaba Cloud in 10 Minutes – A No‑Code Guide

Step 1: Find a solution and start a free trial

OpenAI released two open‑source models, gpt‑oss‑120b and gpt‑oss‑20b, which perform strongly on tool use, few‑shot function calling, CoT reasoning, and health benchmarks. To try them without hardware constraints, use Alibaba Cloud’s PAI Model Gallery for a one‑click, no‑code deployment.

Step 2: Deploy gpt‑oss‑20b model

1. Log in to the PAI console and select Workspace List → New Workspace . Set a workspace name, authorize the service role, and confirm the member addition.

Set the workspace name.

Click “Authorize” in the service role section, confirm, then proceed.

If no members are added, simply confirm to add the current user.

2. Open the model detail page for gpt-oss-20b and click the Deploy button.

3. In the deployment panel, choose llama.cpp Deploy > Single‑Machine Standard and select the resource specification ml.gu7i.c8m30.1-gu30. Keep other parameters at their defaults.

4. If the desired resource is unavailable, switch to another region in the console.

5. Click the Deploy button at the bottom left, confirm the billing reminder, and wait about five minutes for the deployment to finish.

6. After deployment, view the call information, copy the public endpoint and token.

Step 3: Use Cherry Studio client for conversation

Cherry Studio is a mainstream large‑model chat client with MCP integration. Install the client from the official download page or GitHub releases, then configure the model service:

In Settings → Model Service, add a new provider named “PAI” of type OpenAI.

Enter the token obtained earlier as the API key and paste the public endpoint as the API address.

Add the model ID gpt-oss-20b (case‑sensitive) and save.

Test connectivity by clicking the “Detect” button.

Finally, return to the chat interface, select the deployed model, and start a conversation (e.g., ask “Which is larger, 9.11 or 9.9?”).