glm5.1 我配置如下，调用工具一直报错,换其他模型就没事，是这个配置有问题吗？用的英伟达的api 接在new-api 中调用 
 Stream ended without finish_reason  

    "glm": {
      "baseUrl": "https://url/v1",
      "api": "openai-completions",
      "apiKey": "sk-",
      "headers": {
        "User-Agent": "MyCustomClient/1.0"
      },
      "compat": {
        "supportsDeveloperRole": false,
        "requiresReasoningContentOnAssistantMessages": true,...