How to beginner to intermediate · 3 min read

How to iterate and improve prompts

Quick answer

To iterate and improve prompts, start by testing a clear, simple prompt with a model like gpt-4o. Analyze the output, then refine the prompt by adding constraints, examples, or clarifications. Repeat this cycle, comparing outputs to optimize for accuracy and relevance.

PREREQUISITES

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

Setup

Install the OpenAI Python SDK and set your API key as an environment variable to securely authenticate requests.

bash

pip install openai>=1.0

Step by step

Use the OpenAI gpt-4o model to send an initial prompt, then refine it based on the output. This example shows a simple prompt and an improved version with added instructions.

python

import os
from openai import OpenAI

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

# Initial prompt
initial_prompt = "Explain the benefits of exercise."
response = client.chat.completions.create(
    model="gpt-4o",
    messages=[{"role": "user", "content": initial_prompt}]
)
print("Initial output:\n", response.choices[0].message.content)

# Improved prompt with constraints
improved_prompt = (
    "Explain the benefits of exercise in 3 bullet points, "
    "each point no longer than 20 words."
)
response = client.chat.completions.create(
    model="gpt-4o",
    messages=[{"role": "user", "content": improved_prompt}]
)
print("\nImproved output:\n", response.choices[0].message.content)

output

Initial output:
 Exercise improves cardiovascular health, boosts mood, increases energy, and helps maintain a healthy weight.

Improved output:
 - Enhances heart and lung function.
 - Boosts mental health and reduces stress.
 - Supports weight management and muscle strength.

Common variations

You can iterate prompts asynchronously, use streaming for real-time output, or switch models like claude-3-5-sonnet-20241022 for different styles. Adjust max tokens and temperature to control response length and creativity.

python

import os
import asyncio
from openai import OpenAI

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

async def async_prompt():
    response = await client.chat.completions.acreate(
        model="gpt-4o",
        messages=[{"role": "user", "content": "List 3 tips to improve prompt quality."}]
    )
    print(response.choices[0].message.content)

asyncio.run(async_prompt())

output

1. Be specific and clear in your instructions.
2. Use examples to guide the model.
3. Iterate by testing and refining prompts.

Troubleshooting

If the output is too vague or off-topic, add explicit instructions or examples. If the response is too long, limit max tokens or specify length constraints. For inconsistent answers, lower temperature or use few-shot prompting.

✅

Key Takeaways

Start with a clear, simple prompt and analyze the output carefully.
Add constraints, examples, or formatting instructions to guide the model.
Use iterative testing to compare outputs and refine prompts effectively.

Verified 2026-04 · gpt-4o, claude-3-5-sonnet-20241022

Verify ↗