GPT-4 function_call took more than 60 seconds

When using langChain to invoke GPT-4 and formatting the given content as per the specified format, if the response time exceeds 60 seconds, are there any good solutions?

const Recipel = z.object({
      hospital: z.string().describe("医院名称"),
      department: z.string().describe("科室名称"),
      diagnosticPatterns: z.array(z.string().describe("中医诊断的证型")).describe("中医诊断结果"),
      prescriptions: z
         .array(
            z.object({
               chineseMedicineName: z.string().describe("中药名称"),
               dosage: z.string().describe("中药剂量"),
            })
         )
         .describe("中药处方"),
      administrationMethod: z.string().describe("用法,例如:水煎服等"),
      dailyFrequency: z.string().describe("一日几次"),
      quantityPerDose: z.string().describe("每次几付"),
      totalQuantity: z.string().describe("总数量,应以'付'结尾"),
   });

AWS Lambda

export const handler = async (event) => {
   const { CHAT_OPENAI } = await initClient();
   const { input } = event.Payload;

   console.time("extractData execution time");
   const extracts = await extractData({ CHAT_OPENAI, input });
   console.timeEnd("extractData execution time");

   console.log("extractData", extracts);
   return extracts;
};

async function extractData({ CHAT_OPENAI, input }) {
   try {
      const extractionFunctionZodSchema = chooseTemplate();

      const model = CHAT_OPENAI.bind({
         functions: [
            {
               name: "extractor",
               description: `作为资深医疗专业人员,您的职责是将用户提供的中医处方笺内容按照指定格式输出`,
               parameters: zodToJsonSchema(extractionFunctionZodSchema),
            },
         ],
         function_call: { name: "extractor" },
      });

      const results = await model.invoke([new HumanMessage(input)]);

      const {
         additional_kwargs: {
            function_call: { arguments: response },
         },
      } = results;

      return JSON.parse(response);
   } catch (error) {
      console.log("error", error);
      return error;
   }
}