How to beginner · 3 min read

DSPy teleprompter explained

Q: DSPy teleprompter explained

The teleprompter feature in dspy is a utility that streams AI-generated text incrementally, ideal for teleprompter-style applications. It allows you to receive partial outputs as they are generated, enabling smooth real-time display of text from models like openai/gpt-4o-mini.

Quick answer

The teleprompter feature in dspy is a utility that streams AI-generated text incrementally, ideal for teleprompter-style applications. It allows you to receive partial outputs as they are generated, enabling smooth real-time display of text from models like openai/gpt-4o-mini.

PREREQUISITES

Python 3.8+
OpenAI API key (free tier works)
pip install dspy openai>=1.0

Setup

Install dspy and the openai package to use DSPy with OpenAI models. Set your OpenAI API key as an environment variable for secure authentication.

bash

pip install dspy openai>=1.0

Step by step

This example demonstrates how to use dspy to create a teleprompter that streams text from the openai/gpt-4o-mini model. The teleprompter method yields partial outputs as the model generates them.

python

import os
import dspy
from openai import OpenAI

# Initialize OpenAI client
client = OpenAI(api_key=os.environ["OPENAI_API_KEY"])

# Create DSPy LM wrapper
lm = dspy.LM("openai/gpt-4o-mini", client=client)
dspy.configure(lm=lm)

# Use teleprompter to stream output
for partial in dspy.teleprompter("Write a short poem about spring."):
    print(partial, end='', flush=True)

print()  # Newline after streaming completes

output

Write a short poem about spring.

The sun awakens, birds take flight,
Blossoms bloom in colors bright,
Gentle breezes softly sing,
Welcoming the warmth of spring.

Common variations

Use different models by changing the dspy.LM model string, e.g., "openai/gpt-4o".
Run asynchronously with async for loops if your environment supports it.
Customize prompt input or integrate with GUI frameworks to display teleprompter text live.

python

import asyncio

async def async_teleprompter():
    async for partial in dspy.teleprompter("Tell a story about a robot."):
        print(partial, end='', flush=True)

asyncio.run(async_teleprompter())

output

Tell a story about a robot.

Once upon a time, a curious robot named Ada explored the world beyond her factory walls...

Troubleshooting

If streaming output stalls, check your internet connection and API key validity.
Ensure dspy and openai packages are up to date to avoid compatibility issues.
For partial output delays, verify your environment supports asynchronous iteration if using async mode.

✅

Key Takeaways

Use dspy.teleprompter to stream AI-generated text incrementally for teleprompter apps.
Configure dspy.LM with your OpenAI client and model for seamless integration.
Async streaming is supported for responsive UI or CLI teleprompter implementations.

Verified 2026-04 · openai/gpt-4o-mini

Verify ↗