It's a problem for us. When they nerf the model our productivity really drops (due to the specific use case and process we have). We've started using OpenAI and will gradually use it more and more if this continues.