我记得以前gemini-2.0-flash-preview-image-generation这个模型用gemini balance加个图床是可以生图的，现在换成GPT-LOAD之后用这个模型生图报错,具体信息在下面。有没有佬知道这个怎么解决？:tieba_087: 
{
  "message": "400 The requested combination of response modalities (TEXT) is not supported by the model. models/gemini-2.0-flash-preview-image-generation accepts the following combination of response modalities:\n* TEXT, IMAGE\n (request id: 2025090921444970179384...