What is Alibaba Cloud Model Studio - Alibaba Cloud Model Studio

Alibaba Cloud Model Studio integrates the Qwen series and mainstream third-party LLMs with both official Qwen APIs and OpenAI-compatible APIs. Its model services cover multiple modalities, including text, image, audio, and video, and support various scenarios, such as code generation, translation, data mining, and intention recognition. You can use model capabilities on demand without managing the underlying infrastructure, reducing technical O&M burden.

api Call APIs

You can chat with LLMs to perform tasks, such as content creation and summary generation, with just a few lines of code.

Model Studio is compatible with OpenAI API standards. To migrate your existing OpenAI code to Model Studio, you only need to adjust the API key, base URL, and model name.

import os
from openai import OpenAI

# Note: The base URLs for different regions are not interchangeable. The example below uses the base URL for the Singapore region.
# - Singapore: https://dashscope-intl.aliyuncs.com/compatible-mode/v1
# - US (Virginia): https://dashscope-us.aliyuncs.com/compatible-mode/v1
# - China (Beijing): https://dashscope.aliyuncs.com/compatible-mode/v1
client = OpenAI(
    api_key=os.getenv("DASHSCOPE_API_KEY"), 
    base_url="https://dashscope-intl.aliyuncs.com/compatible-mode/v1"
)
completion = client.chat.completions.create(
    model="qwen-plus",
    messages=[{"role": "user", "content": "Who are you?"}]
)
print(completion.choices[0].message.content)

Model service

Model Studio provides out-of-the-box model services. You can directly call the proprietary Qwen models and third-party models, such as DeepSeek and Kimi, without deploying or maintaining them. See Model list.

Qwen flagship models:
- Qwen-Max: The best-performing model in the Qwen series, suitable for handling complex, multi-step tasks.
- Qwen-Plus: A balanced choice for performance, speed, and cost, suitable for general-purpose scenarios.
- Qwen-Flash: Cost-effective and low-latency, suitable for simple tasks that require fast responses.
- Qwen-Coder: Excels at tool calling and environment interaction, and is specialized for code generation and understanding.
Multimodal coverage: Includes various capabilities, such as text generation, visual understanding, image generation, video generation, speech recognition and synthesis, and embedding.
Domain-specific models: Provides various domain models for specific industries and tasks, such as long document processing, translation, data mining, intention recognition, role-playing, and in-depth research.

Billing

Activating Model Studio is free of charge. You are charged when you call models. See Billable items and Model list.

Free quota for new users

Model Studio provides new users with an exclusive free quota in the Singapore region to try out model calls. After the quota is exhausted, billing automatically switches to the pay-as-you-go method. To avoid unexpected charges, you can turn on the Free quota only feature. The service automatically stops when the quota is depleted. See Free quota for new users.

How to pay

Model calls are automatically charged on an hourly basis. For information about supported payment methods, see Introduction to payment methods.

View bills and usage

Billing details: You can visit the Detailed Bills and Cost Analysis pages.
Call statistics: Approximately one hour after a model call is completed, go to the Model Observation (Singapore), Model Observation (Virginia), or Model Observation (Beijing) page, set the query conditions, and click Monitor in the Actions column for the target model. You can then view statistics for the model, such as call volume, token consumption, and success rate. See Model Monitoring.

Getting started with Model Studio

Try models online: Playground (Singapore), Playground (Virginia), or Playground (Beijing)
Make your first API call to Qwen

FAQ

Q: Is my data secure? Will Model Studio use my data for training?

A: No. Alibaba Cloud strictly protects data privacy and will never use your data for model training. The data you transmit when you build applications or train models is encrypted to ensure data security. See Privacy policy.

Q: What are the differences between the Singapore, US (Virginia), and China (Beijing) regions?

A: Alibaba Cloud Model Studio provides model services in the Singapore, US (Virginia), and China (Beijing) regions. Calling a model from a nearby region reduces network latency. Service endpoints (Endpoint/Base URL) and API keys are region-specific and are not interchangeable. The supported models, platform features, and prices also vary by region. See Model list.

Q: How can I avoid automatic payments?

A: Model Studio uses a pay-as-you-go model and does not have an "automatic payment" switch. To avoid charges, you can take the following measures:

Delete your API keys
Go to the API-KEY (Singapore), API-KEY (Virginia), or API-KEY (Beijing) page and delete all API keys. After they are deleted, you can no longer call Model Studio models through APIs, which prevents any further charges.
Stop all model call activities
- Stop model calls in your applications.
- Stop calls from applications such as agents and workflows.
- Check and stop scheduled tasks or background processes.
Charges are triggered by actual calls. Ensure that all call activities have stopped.
Enable the Free quota only feature (for new users only)
On the product page of a supported model, turn on this switch. When the free quota is exhausted, the service stops automatically and returns the error code `AllocationQuota.FreeTierOnly` to prevent charges.
Note: This feature applies only to models in the Singapore region and is valid only during the free quota period.
Set cost monitoring and alerts
- View Detailed Bills.
- Go to Model Observation (Singapore), Model Observation (Virginia), or Model Observation (Beijing) to view call statistics, which are updated about an hour after each call.
- Set up High-spending Alerts. When the daily bill for a specified product exceeds the threshold, the system sends a daily text message reminder so you can take timely action.

By taking these measures, you can effectively control your usage costs.

Q: How to use Qwen3 or DeepSeek models?

Try online: Go to the Models (Singapore), Models (Virginia), or Models (Beijing) page, and click a model to try it out. Note that DeepSeek is supported only in the China (Beijing) region.
Call models through an API: For the procedure, see Make your first API call to Qwen. For a list of supported models, see Model list.
Call models through developer tools (such as Claude Code): See Clients and developer tools.