Function Calling（ツール呼び出し）

モデルに「呼び出すべき関数」を提示し、引数を生成させます。OpenAI の Tool Calling 仕様と互換です。 フローは ①ツール定義つきでリクエスト → ②モデルが tool_calls を返す → ③関数を実行 → ④結果を role: tool で渡して再リクエスト の2往復です。

Python

import os, json
from openai import OpenAI

client = OpenAI(
    api_key=os.environ["LYKURO_API_KEY"],
    base_url="https://api.lykuro.ai/alibaba/compatible-mode/v1",
)

tools = [{
    "type": "function",
    "function": {
        "name": "get_weather",
        "description": "指定した都市の現在の天気を取得する",
        "parameters": {
            "type": "object",
            "properties": {"city": {"type": "string", "description": "都市名"}},
            "required": ["city"],
        },
    },
}]

def get_weather(city: str) -> str:
    return json.dumps({"city": city, "weather": "晴れ", "temp_c": 22})

messages = [{"role": "user", "content": "東京の天気は？"}]

# ① ツール定義つきでリクエスト
resp = client.chat.completions.create(model="qwen-plus", messages=messages, tools=tools)
msg = resp.choices[0].message
messages.append(msg)

# ② モデルが返した tool_calls を実行
for call in msg.tool_calls or []:
    args = json.loads(call.function.arguments)
    result = get_weather(**args)
    messages.append({"role": "tool", "tool_call_id": call.id, "content": result})

# ③ 結果を渡して最終回答を生成
final = client.chat.completions.create(model="qwen-plus", messages=messages, tools=tools)
print(final.choices[0].message.content)

TypeScript

import OpenAI from "openai";

const client = new OpenAI({
  apiKey: process.env.LYKURO_API_KEY,
  baseURL: "https://api.lykuro.ai/alibaba/compatible-mode/v1",
});

const tools: OpenAI.Chat.ChatCompletionTool[] = [{
  type: "function",
  function: {
    name: "get_weather",
    description: "指定した都市の現在の天気を取得する",
    parameters: {
      type: "object",
      properties: { city: { type: "string", description: "都市名" } },
      required: ["city"],
    },
  },
}];

const getWeather = (city: string) =>
  JSON.stringify({ city, weather: "晴れ", temp_c: 22 });

const messages: OpenAI.Chat.ChatCompletionMessageParam[] = [
  { role: "user", content: "東京の天気は？" },
];

const resp = await client.chat.completions.create({ model: "qwen-plus", messages, tools });
const msg = resp.choices[0].message;
messages.push(msg);

for (const call of msg.tool_calls ?? []) {
  const args = JSON.parse(call.function.arguments);
  messages.push({ role: "tool", tool_call_id: call.id, content: getWeather(args.city) });
}

const final = await client.chat.completions.create({ model: "qwen-plus", messages, tools });
console.log(final.choices[0].message.content);

Go

package main

import (
	"context"
	"encoding/json"
	"fmt"
	"os"

	openai "github.com/sashabaranov/go-openai"
)

func getWeather(city string) string {
	b, _ := json.Marshal(map[string]any{"city": city, "weather": "晴れ", "temp_c": 22})
	return string(b)
}

func main() {
	cfg := openai.DefaultConfig(os.Getenv("LYKURO_API_KEY"))
	cfg.BaseURL = "https://api.lykuro.ai/alibaba/compatible-mode/v1"
	client := openai.NewClientWithConfig(cfg)

	tools := []openai.Tool{{
		Type: openai.ToolTypeFunction,
		Function: &openai.FunctionDefinition{
			Name:        "get_weather",
			Description: "指定した都市の現在の天気を取得する",
			Parameters: map[string]any{
				"type":       "object",
				"properties": map[string]any{"city": map[string]any{"type": "string"}},
				"required":   []string{"city"},
			},
		},
	}}

	messages := []openai.ChatCompletionMessage{{Role: "user", Content: "東京の天気は？"}}
	ctx := context.Background()

	resp, err := client.CreateChatCompletion(ctx, openai.ChatCompletionRequest{
		Model: "qwen-plus", Messages: messages, Tools: tools,
	})
	if err != nil {
		panic(err)
	}
	msg := resp.Choices[0].Message
	messages = append(messages, msg)

	for _, call := range msg.ToolCalls {
		var args struct {
			City string `json:"city"`
		}
		_ = json.Unmarshal([]byte(call.Function.Arguments), &args)
		messages = append(messages, openai.ChatCompletionMessage{
			Role: openai.ChatMessageRoleTool, ToolCallID: call.ID, Content: getWeather(args.City),
		})
	}

	final, err := client.CreateChatCompletion(ctx, openai.ChatCompletionRequest{
		Model: "qwen-plus", Messages: messages, Tools: tools,
	})
	if err != nil {
		panic(err)
	}
	fmt.Println(final.Choices[0].Message.Content)
}

cURL

# ① ツール定義つきでリクエスト（モデルが tool_calls を返す）
curl https://api.lykuro.ai/alibaba/compatible-mode/v1/chat/completions \
  -H "Authorization: Bearer $LYKURO_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{
    "model": "qwen-plus",
    "messages": [{"role": "user", "content": "東京の天気は？"}],
    "tools": [{
      "type": "function",
      "function": {
        "name": "get_weather",
        "description": "指定した都市の現在の天気を取得する",
        "parameters": {
          "type": "object",
          "properties": {"city": {"type": "string"}},
          "required": ["city"]
        }
      }
    }]
  }'

返ってきた tool_calls の関数をアプリ側で実行し、その結果を {"role":"tool","tool_call_id":"...","content":"..."} として messages に追加し、もう一度同じエンドポイントへ送ると最終回答が得られます。

:::tip 対応モデル Function Calling は qwen-plus / qwen-max / qwen3-max / deepseek-chat などで利用できます。 対応状況は モデル一覧 を参照してください。 :::