Qwen Code

Use a Valkyrie model in the Qwen Code CLI over the OpenAI-compatible API.

Qwen Code is an OpenAI-compatible coding CLI. You can point it at a Valkyrie deployment with a few environment variables, or with a settings file if you want it saved.

Quick setup with environment variables

export OPENAI_API_KEY="vk_dep_your_deployment_key"
export OPENAI_BASE_URL="https://valkyrie-back.azumo.com/{deployment_slug}/v1"
export OPENAI_MODEL="your-model-id"

qwen

Environment variables take priority. You can also put them in ~/.qwen/.env (or ./.qwen/.env for a single project).

Saved setup with settings.json

To keep the config, add a provider to ~/.qwen/settings.json:

{
  "modelProviders": {
    "openai": [
      {
        "id": "RadixArk/Qwen3.8-27B-NVFP4",
        "name": "Valkyrie Qwen 27B",
        "envKey": "VALKYRIE_API_KEY",
        "baseUrl": "https://valkyrie-back.azumo.com/qwen/v1",
        "generationConfig": {
          "contextWindowSize": 131072,
          "samplingParams": { "max_tokens": 16384 }
        }
      }
    ]
  },
  "security": { "auth": { "selectedType": "openai" } },
  "model": { "name": "RadixArk/Qwen3.8-27B-NVFP4" }
}

envKey names the environment variable that holds your key, so the key itself stays out of the file.

The base URL ends in /v1. Use your deployment slug (deployment-{id}) or an alias, and set the model id to what your deployment serves.