IMO you’re better off using qwen3.5-plus through the model studio coding plan, but ymmv