RT，无论是在cherry还是CC里面，所有的回答（哪怕是比较复杂的问题）都是秒出，看不到思维链，这是我目前的config 
    override:
      - models:
          - name: "gemini-claude-opus-4-6-thinking"
            protocol: "antigravity"
        params:
          "generationConfig.maxOutputTokens": 32000
          "generationConfig.thinkingConfig.thinkingBudget": 31999
          "generationConfig.thinkingConfig.includeThoughts": true
oauth-model-alias:
...