How to Stream Tokens from a Mock LLM in Python

Simulate real-time LLM streaming by yielding tokens one at a time with a delay, making it easy to test streaming UIs.

Easy Python 3.9+ Aug 9, 2026 AI & LLM integration patterns 15 views 0 copies

generator llm streaming simulation typing

Python code

18 lines

Python 3.9+

import time
from typing import Generator


def stream_tokens(text: str, delay: float = 0.05) -> Generator[str, None, None]:
    """Simulate an LLM streaming tokens word by word."""
    for word in text.split():
        yield word
        time.sleep(delay)


if __name__ == "__main__":
    sample = "Hello world! This is a simulated streaming output."
    collected = []
    for token in stream_tokens(sample):
        print(token, end=" ", flush=True)
        collected.append(token)
    print("\nCollected tokens:", collected)

Output

stdout

Hello world! This is a simulated streaming output. 
Collected tokens: ['Hello', 'world!', 'This', 'is', 'a', 'simulated', 'streaming', 'output.']

How it works

The stream_tokens function is a generator — it uses yield instead of return, so each call to next() (or iteration) resumes execution right after the last yield, producing one word at a time. The time.sleep adds a realistic delay to mimic network latency, making the streaming visible in the output. The typing.Generator annotation clarifies the return type to static checkers and readers.

Common mistakes

Using `return` instead of `yield`, which turns the function into a regular function and breaks the iteration.
Forgetting `flush=True` in `print`, causing output to appear all at once instead of incrementally.
Calling the generator function without iterating over it (must use `for` loop or `next()`).

Variations

Use `yield from` with a sub-generator to chunk longer text into fixed-size batches.
Yield full lines or sentences instead of words by splitting on `.` or `\n`.

Real-world use cases

Testing a chat UI in development without a live LLM API — iterate on the streaming display logic locally.
Demoing token-by-token streaming at a tech talk or internal showcase where a real API call is impractical.
Stress-testing a frontend that renders partial model responses by varying the delay parameter.

How to Stream Tokens from a Mock LLM in Python

Python code

Output

How it works

Common mistakes

Variations

Real-world use cases

More from AI & LLM integration patterns

Tutorials

Quizzes

Python code

Output

How it works

Common mistakes

Variations

Real-world use cases

More from AI & LLM integration patterns

Keep learning

Tutorials

Quizzes