DeepSeek V4 Flash 正式版上线：附完整实测

Input length 1280009 exceeds the maximum length 1048570

import os
from openai import OpenAI

client = OpenAI(
    api_key=os.environ["APIYI_API_KEY"],
    base_url="https://api.apiyi.com/v1",
)

# 简单任务：显式关思考，实测可省掉 200+ 思考 token
resp = client.chat.completions.create(
    model="deepseek-v4-flash-ga-260731",
    messages=[{"role": "user", "content": "把这段话改写得更简洁：……"}],
    extra_body={"thinking": {"type": "disabled"}},
)
print(resp.choices[0].message.content)

resp = client.chat.completions.create(
    model="deepseek-v4-flash-ga-260731",
    messages=[{"role": "user", "content": "分析这段代码的时间复杂度并给出优化方案"}],
    extra_body={"thinking": {"type": "enabled"}},
)
msg = resp.choices[0].message
print("思考过程：", getattr(msg, "reasoning_content", None))
print("最终回答：", msg.content)

tools = [{
    "type": "function",
    "function": {
        "name": "submit_result",
        "description": "提交抽取结果",
        "parameters": {
            "type": "object",
            "properties": {
                "city": {"type": "string"},
                "temp_c": {"type": "number"},
            },
            "required": ["city", "temp_c"],
        },
    },
}]

resp = client.chat.completions.create(
    model="deepseek-v4-flash-ga-260731",
    messages=[{"role": "user", "content": "北京今天 25 度"}],
    tools=tools,
)
# 工具参数是真正被约束的，可直接解析
import json
args = json.loads(resp.choices[0].message.tool_calls[0].function.arguments)
print(args)