Pythonstreaming_demo.py
Copy
1
import time
2
from openai import OpenAI
3
 
4
client = OpenAI()
5
 
6
def stream_chat(prompt: str) -> str:
7
    start = time.monotonic()
8
    first_token_time = None
9
    full_response = ""
10
 
11
    stream = client.chat.completions.create(
12
        model="gpt-4o-mini",
13
        messages=[{"role": "user", "content": prompt}],
14
        stream=True,
15
    )
16
 
17
    for chunk in stream:
18
        delta = chunk.choices[0].delta.content
19
        if delta:
20
            if first_token_time is None:
21
                first_token_time = time.monotonic()
22
                ttft = first_token_time - start
23
                print(f"\n[TTFT: {ttft:.3f}s]\n")
24
            print(delta, end="", flush=True)
25
            full_response += delta
26
 
27
    total = time.monotonic() - start
28
    print(f"\n\n[Total: {total:.3f}s | {len(full_response)} chars]")
29
    return full_response
30
 
31
if __name__ == "__main__":
32
    stream_chat("Explain quantum entanglement in 3 paragraphs.")
33