Alibaba Cloud Model Studio:FAQ

Common errors and solutions

What should I do if Claude Code shows “Claude Code has switched from npm to native installer”?

This message does not affect normal use of Claude Code. Run claude install in your terminal to migrate to the official native installer, and follow the commands returned in the terminal to complete the configuration migration.

What should I do if OpenCode shows "Request Entity Too Large"?

This error indicates that the request content, such as code context or conversation history, is too large and exceeds the maximum input limit. To resolve this issue:

1. Run /new to start a new conversation, or run /compact to compress the context. For more ways, see What is the context window for each model? What to do if I exceed the context window?.

2. If the preceding methods do not resolve the issue, update OpenCode to version 1.2.16 or later, which includes a fix for this problem.

Why does OpenClaw show call records of models that I never configured?

Cause: The openclaw.json file does not have agents.defaults.models configured to restrict the allowed model list.

Solution: Explicitly declare the list of allowed models in the agents.defaults.models field of ~/.openclaw/openclaw.json. For details, see Set up Coding Plan.

What should I do if OpenClaw shows as connected but the chat has no response or keeps loading?

Possible causes:

1. A local proxy (VPN or HTTP proxy) is intercepting or failing to forward requests to coding.dashscope.aliyuncs.com.

2. The base URL in the OpenClaw cache file ~/.openclaw/agents/main/agent/models.json points to an incorrect address, causing requests to be sent to an unresponsive endpoint.

Solution:

1. Disable your local proxy, or make sure that the proxy can properly access coding.dashscope.aliyuncs.com.

2. Delete the ~/.openclaw/agents/main/agent/models.json cache file and restart OpenClaw. After the restart, the cache file is automatically rebuilt based on the configuration in ~/.openclaw/openclaw.json.

What should I do if OpenClaw shows "Agent failed before reply: OAuth token refresh failed"?

This error indicates that a third-party OAuth service is connected instead of Coding Plan.

Solution:

1. Set up Coding Plan in OpenClaw. For details, see Set up Coding Plan.

2. Clear the old OAuth credentials and model cache, and then restart OpenClaw.

   Before deleting old OAuth credentials, back up the related files so that you can roll back if needed.

     rm ~/.openclaw/agents/main/agent/auth.json
     rm ~/.openclaw/agents/main/agent/models.json
     openclaw gateway restart

What should I do if OpenClaw shows “API rate limit reached”?

This error indicates that the quota of your current Coding Plan subscription has been reached, or a configuration issue caused the request to be rejected.

Troubleshoot in the following order:

1. Check whether you have exceeded your quota: Go to the Coding Plan pageCoding Plan page to view your usage.

   1. If your quota is exhausted, check the next quota reset time on the page.

   2. If you frequently reach the limit, consider upgrading to Pro.

2. Check for OpenClaw configuration errors: If your quota is not exceeded, carefully verify the configuration file against . Make sure that the models, agents, and gateway fields (including nested fields) match the documentation to avoid request rejections caused by configuration errors.

3. Reset your API key: If the issue persists after the preceding checks, go to the Coding Plan pageCoding Plan page to reset your API key.

Why do I get "HTTP 401: Incorrect API key provided."?

Possible reasons:

• Incorrect API key format: The API key is empty, formatted incorrectly, copied incompletely, or contains extra spaces. Make sure that you are using an exclusive Coding Plan API Key (starting with sk-sp-), it is copied completely, and has no extra spaces.

• Subscription expired or invalid: If your subscription has expired or is invalid, the exclusive API key will not work. Make sure that your subscription is active.

• Incorrect base URL used: You have configured an exclusive Coding Plan API key (starting with sk-sp-), but the base URL is still the general one, such as https://dashscope-intl.aliyuncs.com/compatible-mode/v1. Replace the base URL with an exclusive URL from the following table, based on the AI tool you are using.

  Tool	Protocol	Base URL
  OpenClaw	OpenAI compatible	https://coding-intl.dashscope.aliyuncs.com/v1
  OpenCode	Anthropic compatible	https://coding-intl.dashscope.aliyuncs.com/apps/anthropic/v1
  Claude Code	Anthropic compatible	https://coding-intl.dashscope.aliyuncs.com/apps/anthropic
  Cursor	OpenAI compatible	https://coding-intl.dashscope.aliyuncs.com/v1
  VSCode Cline	OpenAI compatible	https://coding-intl.dashscope.aliyuncs.com/v1
  Qwen Code	OpenAI compatible	https://coding-intl.dashscope.aliyuncs.com/v1

• Incorrect API key used: You have configured the exclusive base URL but are using a general API key (starting with sk-). Use an exclusive Coding Plan API Key.

• Cached OpenClaw setting: Delete the providers.bailian item from ~/.openclaw/agents/main/agent/models.json, and restart OpenClaw.

Why am I still charged for model API calls after subscribing to Coding Plan?

Possible reasons:

1. If you configured a general-purpose API key (format: sk-xxxxx) and a general-purpose base URL (without the keyword "coding") in your AI tool, the system treats these calls as pay-as-you-go requests. These calls cannot consume your Coding Plan quota and will generate separate charges. Configure the correct subscription-specific credentials. For details, see Get exclusive API key and base URL.

2. The charge may result from API calls made before you activated Coding Plan. Billing is processed hourly, and during peak periods, there may be a delay of up to one hour. For example, if you called the API at 16:00 and activated Coding Plan at 19:00, a bill generated at 19:30 would reflect the 16:00 usage. The actual billing time determines the charge. For details, see Bill inquiry and cost management.

What to do when my quota runs out?

• Lite: After using up your 5-hour or weekly quota, wait for automatic restoration. After using up your monthly quota, upgrade to Pro or wait until your next subscription month begins and your quota renews.

• Pro: After using up your 5-hour or weekly quota, wait for automatic restoration. After using up your monthly quota, wait until your next subscription month begins and your quota renews.

Will my usage switch to pay-as-you-go after my Coding Plan quota is used up?

No. After your Coding Plan quota is used up, further API calls will fail with an error. Usage will not automatically switch to pay-as-you-go billing. To continue using the service, upgrade to Pro for more quota or wait for your quota to refresh in the next subscription cycle.

Can Coding Plan use free trial models from Model Studio?

No. Coding Plan is a standalone subscription product. Its billing and quota system does not participate in Model Studio's free quota program.

If I buy Coding Plan without enabling auto-renewal, can I still get the 50% discount on my first manual renewal next month?

Yes. Your first manual renewal still qualifies for the 50% discount.

Why does my Coding Plan subscription purchased in February last only 28 days instead of 31?

Your subscription starts at the moment of activation and expires at 23:59:59 (UTC+8) on the same date of the following month. If the following month has no corresponding date, it expires at 23:59:59 (UTC+8) on the last day of that month.

Example: If you activate Coding Plan on February 3, 2026, it expires on March 3, 2026, resulting in 28 days of remaining validity.

How can I view token consumption details?

This is currently not supported. Coding Plan quota consumption is based on the number of model calls, not token usage. You can view your usage on the Coding Plan page.

Does Coding Plan have an annual plan?

Coding Plan is available only on a monthly subscription basis. Annual plans are currently not available.

How many Coding Plan subscriptions can one account have at the same time?

Each Model Studio account can subscribe to only one Coding Plan at a time (regardless of Lite or Pro).

Can I use models not listed in the supported models list?

Coding Plan supports only the models listed in the supported model list. If you use any other model, an error occurs.

Is the response speed the same for Lite and Pro?

The response speed is identical for Lite and Pro because they use the same model resources and inference services.

What is the context window for each model? What to do if I exceed the context window?

If you receive an error about exceeding the context limit, start a new session. To prevent this error:

1. Switch models: Use a model with a longer context window, such as qwen3.5-plus or qwen3-coder-plus.

2. Reduce unnecessary files: Start the AI coding tool in a specific project folder with only the files you need.

3. Split tasks: Break complex tasks into smaller subtasks and ask one question at a time to reduce context usage per request.

4. Use precise instructions: Vague requests may trigger unnecessary file scans. Ask clear, specific questions or give exact instructions.

Exact steps vary by tool, see Set up AI tools.

How to reset the API key for Coding Plan?

On the Coding Plan page, click Reset next to the API key.

If my Coding Plan expires and I resubscribe, will my API key be reset?

It will be reset unless you renew it before expiration.

Can I generate multiple API keys for Coding Plan?

No. Only one API key is supported.

How can a RAM user view or get the Coding Plan API key?

Contact your Alibaba Cloud account owner or a RAM user with administrative permission to grant your account the Subscription Plans permission.

How do I use Coding Plan in VS Code?

Use extensions that support OpenAI- or Anthropic-compatible APIs, such as Qwen Code or Claude Code.

Do Coding Plan models support thinking mode?

Most Coding Plan models that support deep thinking also support thinking mode by default. To enable it, ensure your selected model supports thinking mode. Common methods include the following:

The model you use must support thinking mode in the first place.

{
  "thinking": {
    "type": "enabled",
    "budgetTokens": 1024
  }
}

{
  "ide": {
    "hasSeenNudge": true
  },
  "env": {
    "BAILIAN_CODING_PLAN_API_KEY": "sk-sp-xxx"
  },
  "modelProviders": {
    "openai": [
      {
        "id": "qwen3.5-plus",
        "name": "[Bailian Coding Plan] qwen3.5-plus",
        "baseUrl": "https://coding-intl.dashscope.aliyuncs.com/v1",
        "envKey": "BAILIAN_CODING_PLAN_API_KEY",
        "generationConfig": {
          "extra_body": {
            "enable_thinking": true
          }
        }
      },
      ...
    ]
  },
  "security": {
    "auth": {
      "selectedType": "openai"
    }
  },
  "codingPlan": {
    "region": "china",
    "version": "xxx"
  },
  "model": {
    "name": "qwen3.5-plus"
  },
  "$version": 3
}

Is the model in Coding Plan quantized?

Coding Plan uses open-source models deployed directly without quantization.

How do I handle the data_inspection_failed error?

See the Error messages documentation.

Why does Coding Plan fail when configured in Dify?

Do not use Coding Plan in Dify. Coding Plan quota is intended only for use in programming tools (such as Claude Code or Qwen Code). It is prohibited to use the subscription API key in automated scripts, custom application backends, or any non-interactive batch calling scenarios. Such use is considered a violation and may result in your subscription being suspended or your API key being blocked.

Can I call Coding Plan using Postman?

The Coding Plan quota is intended only for use in coding tools, such as Claude Code and Qwen Code. You are prohibited from using this quota to make API calls for automated scripts, custom application backends, or any non-interactive batch call scenarios. Such use is considered a violation and may result in your subscription being suspended or your API key being blocked.