In a Singapore Model Studio workspace, call OpenAI-compatible Chat Completions with the model ID qwen4-max. Set reasoning_effort to low, high, or xhigh; the default is high.
Suitable tasks: Calling Qwen4-Max on Alibaba Cloud Model Studio through cURL or an OpenAI-compatible client
Unsuitable tasks: Requests that need both a reasoning level and a reasoning-token budget set at the same time
Applicable model version: qwen4-max (Latest Model), as listed on the official model page
Applicable clients, agents, or APIs: OpenAI Chat Completions-compatible API and cURL; for other regions or API formats, replace the regional endpoint or integration protocol as appropriate
Recommended reasoning levels and parameters: Use high (the default) for complex tasks; use low when prioritizing latency or for lighter tasks; use xhigh for greater reasoning intensity
This example is fixed to the Singapore region and a workspace-specific domain. Securely set the WORKSPACE_ID and DASHSCOPE_API_KEY environment variables in the current terminal first. Do not put the key in a script, code repository, or request body:
: "${WORKSPACE_ID:?请设置 Model Studio 新加坡区域的 Workspace ID}"
: "${DASHSCOPE_API_KEY:?请安全设置 Model Studio API Key}"
curl --fail-with-body \
"https://${WORKSPACE_ID}.ap-southeast-1.maas.aliyuncs.com/compatible-mode/v1/chat/completions" \
-H "Authorization: Bearer ${DASHSCOPE_API_KEY}" \
-H "Content-Type: application/json" \
-d '{
"model": "qwen4-max",
"messages": [
{"role": "user", "content": "用三点总结什么是缓存。"}
],
"reasoning_effort": "high"
}'| Configuration item | Value |
|---|---|
| Region | Singapore (新加坡) |
| Base URL | https://{WorkspaceId}.ap-southeast-1.maas.aliyuncs.com/compatible-mode/v1 |
| Chat Completions path | /chat/completions |
| Model ID | qwen4-max |
| Reasoning levels | low, high, xhigh; default high |
You can find the WorkspaceId under Workspace Details in the Model Studio console. The official documentation recommends using a workspace-specific domain for Singapore requests; the legacy domain is still available. Get the API key from the Model Studio console for the corresponding region.
Select the Singapore region in the Model Studio console, confirm the workspace, and obtain the Workspace ID and API key.
Set WORKSPACE_ID and DASHSCOPE_API_KEY in the current shell, then run the cURL request above.
Switch to low or xhigh to adjust reasoning intensity; do not submit thinking_budget at the same time.
The model page lists the model as qwen4-max (Latest Model), gives qwen4-max as the model invocation ID, and lists the supported reasoning_effort levels and default value.
The API documentation's Singapore example gives https://{WorkspaceId}.ap-southeast-1.maas.aliyuncs.com/compatible-mode/v1 and the /chat/completions path, and explains where to find the Workspace ID in the console.
The API documentation states that the default reasoning_effort value for qwen4-max is high, with low, high, and xhigh as valid levels; reasoning_effort and thinking_budget cannot be set at the same time.
This configuration demonstrates only OpenAI-compatible Chat Completions in the Singapore region. For China (Beijing), China (Hong Kong), the US (Virginia), Germany (Frankfurt), and Japan (Tokyo), use the API domain and workspace for the corresponding region.
This article uses the exact model ID currently listed by the official source, qwen4-max, without adding a preview or snapshot suffix; availability and quotas depend on the console for the selected region.
The official model page states that the model supports text, image, and video input; this example sends text only. The model context length is 1,000,000 tokens. The maximum output length varies by thinking mode, and the example does not explicitly set an output limit.
The model and regional endpoint pages were checked in Tabbit on 2026-11-17. The model page shows an update date of 2026-11-16. The official documentation provides the model ID, regional domain, and reasoning levels used here. The API request was untested with a real key. OpenAI-compatible Chat API documentation
Qwen4-Max