Qwen Code CLI Custom Model Configuration

Qwen Code uses the modelProviders configuration in ~/.qwen/settings.json to customize the Provider. Here, the OpenAI-compatible protocol is fixed.

¶ Configuration

{
  "modelProviders": {
    "acedatacloud": [
      {
        "id": "MODEL_ID",
        "name": "MODEL_ID",
        "envKey": "ACEDATACLOUD_API_KEY",
        "baseUrl": "https://api.acedata.cloud/v1",
        "provider": "openai"
      }
    ]
  },
  "providerProtocol": { "acedatacloud": "openai" }
}

modelProviders.acedatacloud must be an array. Provider entries are independently configured and will not automatically inherit other top-level context, Header, or body parameters.

¶ Verification

export ACEDATACLOUD_API_KEY="PASTE_YOUR_ACEDATACLOUD_API_TOKEN_HERE"
qwen --auth-type openai --model MODEL_ID --output-format stream-json \
  --prompt "Reply only OK"

After entering, send Reply only OK, then run the read-only file tool again. If the strictly compatible endpoint rejects the tool result format, please check the current Qwen Code version and the official compatibility instructions.

¶ Official References

• Qwen Code Settings
• Qwen Code Model Providers

Return to the Coding Plan Configuration Center.