Alibaba Qwen

Serve Qwen models from Alibaba Model Studio through your Verdictan gateway. The gateway calls the OpenAI-compatible mode of Model Studio with your Alibaba key, and your application calls the gateway with its Verdictan application token.

This page connects Qwen as a model provider. To route the Qwen Code command-line tool through Verdictan, see Qwen Code CLI.

Before you start

• A working gateway from the quickstart.
• An Alibaba Model Studio API key, and one Qwen model ID that your account can use.
• The gateway startup command, and a separate application token for test requests.

Configure the provider

Use provider: qwen or provider: alibaba. Avoid the dashscope alias: it works for chat, but the gateway does not route embeddings for it.

policy-config.yaml

pack:
  name: qwen-integration
  version: 1.0.0
  enabled: true
policies:
  chain:
    - prompt-injection
providers:
  targets:
    - id: qwen-primary
      provider: qwen
      model: "replace-with-qwen-model-id"
      base_url: https://dashscope-intl.aliyuncs.com/compatible-mode
      secret_key_ref:
        env: VERDICTAN_QWEN_API_KEY

The example uses the international compatible-mode address, which is also the default. If your account belongs to another region, set base_url to that region's address, which must start with https://. The gateway sends the key from VERDICTAN_QWEN_API_KEY as a Bearer token.

Start and verify

You can use CLI 0.1.14 or later for the Qwen configuration deployment. After In sync, send the request below to verify the selected model and provider key.

With the quickstart in Docker, follow the container steps instead of steps 1 to 3.

1. In the gateway terminal, set the key and lint the file:

   export VERDICTAN_QWEN_API_KEY="<provider-api-key>"
   verdictan policy lint --file policy-config.yaml

2. Restart the test gateway with the key in its environment.

3. Deploy the file to the test agent in Configurations. If that agent already runs pack.version 1.0.0, increase the version first. Then wait for the gateway log line connected read model refresh applied updated config with your version as config_version. Try the change on a test gateway has the steps.

4. In a second terminal, send a request for the configured model:

   export VERDICTAN_CLIENT_TOKEN="<application-token>"
   curl --fail-with-body http://127.0.0.1:41002/v1/chat/completions \
     -H "Authorization: Bearer ${VERDICTAN_CLIENT_TOKEN}" \
     -H "Content-Type: application/json" \
     -d '{"model":"replace-with-qwen-model-id","messages":[{"role":"user","content":"Reply with Hello."}]}'

choices[0].message.content contains the reply, and the request appears in History when capture is on, or in the delivered events.

Supported requests

Request	Support
Chat Completions	Yes, including streaming and tool calls
Messages	Translated to Chat Completions
Embeddings	Yes, with provider: qwen or alibaba, and an embedding model. The policy chain does not run on embeddings.
Responses, image input, response_format, parallel tool calls	No

Failure behavior

Problem	What to check
401 from the gateway	The application token is active, and the key is set on the gateway host.
Provider authentication error	VERDICTAN_QWEN_API_KEY is set in the gateway process, and it belongs to the region of base_url.
Model not found	The request model matches the target, and your account can use it.
capability.unsupported_family	The request uses a family from the table above, and the alias is not dashscope for embeddings.
Provider 429 or another upstream error	The response body and your Model Studio quota. See Error Handling.

Limits

• Model names, regional addresses, prices, rate limits, and data terms are in the Alibaba Model Studio documentation. Verdictan does not publish them.
• Streaming and tools vary by Qwen model. Test each one with your model.
• A listener on a non-loopback address needs TLS and network controls.

Next steps

• Provider Configuration: target fields, routing, and fallback.
• Qwen Code CLI: route the Qwen coding tool through the gateway.