I wouldn't say that, but just that qwen 3 max thinking definitely underperforms relative to its size.