Alibaba Cloud Model Studio:Claude Code

Supported models

The Model Studio Anthropic API-compatible service supports the following Qwen models:

For model specifications and billing rules, see the Model list.

Model recommendations

Use the following models for different task types:

• Complex reasoning tasks, such as designing application architecture or implementing complex algorithms: Use Qwen-Plus series models (upgraded to Qwen3.5), Qwen-Max series models, or qwen3-coder-plus.

• Simple coding tasks, such as writing functions or generating code comments: Use Qwen-Flash series models or qwen3-coder-flash.

1. Activate Alibaba Cloud Model Studio

1. Create an account: If you do not have an Alibaba Cloud account, create one.

   If you encounter problems, see Create an Alibaba Cloud account.

2. Activate Alibaba Cloud Model Studio: Use your Alibaba Cloud account to go to Alibaba Cloud Model Studio. After you read and agree to the Terms of Service, Model Studio automatically activates. If the Terms of Service do not appear, the service has already been activated.

   If a message indicates that you have not completed identity verification, complete identity verification first.

After you activate Model Studio for the first time, you receive a free quota for model inference valid for 90 days. For more information, see the Free quota for new users page.

2. Install Claude Code

npm install -g @anthropic-ai/claude-code

3. Configure the base URL, API key, and model

To access the model service in Model Studio using an Anthropic API-compatible method, configure the following environment variables.

1. ANTHROPIC_BASE_URL: Set this to https://dashscope-intl.aliyuncs.com/apps/anthropic.

2. ANTHROPIC_API_KEY or ANTHROPIC_AUTH_TOKEN: Set this to your Alibaba Cloud Model Studio API key.

   ANTHROPIC_API_KEY or ANTHROPIC_AUTH_TOKEN for authentication. This topic uses ANTHROPIC_API_KEY as an example.

3. ANTHROPIC_MODEL: Set this to a supported model from the Model list.

4. Run claude code

Go to the project directory and run the claude command in the terminal.

claude

The first time you use claude code, you might need to log on to an Anthropic account. To skip this step:

1. Locate the .claude.json file in your user home directory:

   • macOS / Linux: ~/.claude.json

   • Windows: C:\Users\%USERNAME%\.claude.json

2. Set the value of the hasCompletedOnboarding field to true.

   {
     "hasCompletedOnboarding": true
   }

3. Save the file, and then run claude again in a new terminal.

(Optional) More ways to configure models

Claude Code supports the following model configuration methods, listed in descending order of priority. A higher-priority configuration overwrites a lower-priority one.

1. During a conversation: Run the /model <model_name> command to switch models. This method is suitable for temporarily switching models.

   /model qwen3.5-plus

2. When starting Claude Code: Run claude --model <model_name> to specify a model. This method is suitable for a single session.

   claude --model qwen3.5-plus

3. Set environment variables: Set environment variables to select different models based on task complexity. Claude Code automatically selects the appropriate model based on the task type. This method is suitable for global configuration.

   export ANTHROPIC_DEFAULT_OPUS_MODEL="qwen3.5-plus"
   export ANTHROPIC_DEFAULT_SONNET_MODEL="qwen3.5-plus"
   export ANTHROPIC_DEFAULT_HAIKU_MODEL="qwen3-coder-flash"

   Where:

   • ANTHROPIC_DEFAULT_OPUS_MODEL: Used for high-difficulty tasks such as complex reasoning and architecture design.

   • ANTHROPIC_DEFAULT_SONNET_MODEL: Used for common tasks such as code writing and feature implementation.

   • ANTHROPIC_DEFAULT_HAIKU_MODEL: Used for simple jobs such as syntax checking and file searching.

4. Set in the settings.json configuration file: Create a settings.json file in the project root directory or user home directory and add the model configuration. This method provides a permanent configuration at the project or user level.

   {
     "env": {
       "ANTHROPIC_DEFAULT_OPUS_MODEL": "qwen3.5-plus",
       "ANTHROPIC_DEFAULT_SONNET_MODEL": "qwen3.5-plus",
       "ANTHROPIC_DEFAULT_HAIKU_MODEL": "qwen3-coder-flash"
     }
   }

Question 1: After starting claude code, the interface displays "Unable to connect to Anthropic services. Failed to connect to api.anthropic.com: ERR_BAD_REQUEST". What should I do?

You can configure the environment variables again and then run the claude command in the terminal.

Question 2: My tokens are consumed too quickly. How do I save tokens?

Claude Code scans the entire project directory, reads relevant code files, and maintains the full conversation history to provide coding suggestions. As a result, token consumption is significantly higher than in typical chat scenarios. Use the following methods to control consumption:

1. Reduce irrelevant files: To avoid consuming tokens on scanning irrelevant files, start Claude Code in a specific project directory and keep only the necessary project files.

2. Summarize conversations: Claude Code uses the conversation history as context. When the conversation length reaches 95% of the context window, Claude Code automatically summarizes the content. You can also run the /compact command to summarize the content manually.

3. Use precise instructions: Vague requests can trigger unnecessary file scans and consume more tokens. Ask clear and specific questions or provide precise instructions when using Claude Code.

4. Break down tasks: When handling complex tasks, break them down into several simple jobs.

5. Reset context: Before starting a new task, use the /clear command to reset the context and prevent irrelevant information from consuming tokens.

For more tips on saving tokens, see the official Claude Code documentation. To view your exact token consumption and expense details, see Bill inquiry and cost management.