How to beginner · 4 min read

How to do zero-shot text classification with OpenAI

Q: How to do zero-shot text classification with OpenAI

Use the OpenAI API's chat.completions.create method with a prompt that instructs the model to classify text into given categories without prior training. Provide the text and candidate labels in the message content for zero-shot classification using models like gpt-4o.

Quick answer

Use the OpenAI API's chat.completions.create method with a prompt that instructs the model to classify text into given categories without prior training. Provide the text and candidate labels in the message content for zero-shot classification using models like gpt-4o.

PREREQUISITES

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

Setup

Install the official OpenAI Python SDK and set your API key as an environment variable.

Install SDK: pip install openai
Set environment variable: export OPENAI_API_KEY='your_api_key' (Linux/macOS) or setx OPENAI_API_KEY "your_api_key" (Windows)

bash

pip install openai

output

Collecting openai
  Downloading openai-1.x.x-py3-none-any.whl (xx kB)
Installing collected packages: openai
Successfully installed openai-1.x.x

Step by step

Use the OpenAI client to send a prompt that asks the model to classify the input text into specified labels. This example uses gpt-4o for zero-shot classification.

python

import os
from openai import OpenAI

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

text_to_classify = "I love the new design of your product!"
candidate_labels = ["positive", "negative", "neutral"]

prompt = (
    f"Classify the following text into one of these categories: {', '.join(candidate_labels)}.\n"
    f"Text: \"{text_to_classify}\"\n"
    "Return only the label."
)

response = client.chat.completions.create(
    model="gpt-4o",
    messages=[{"role": "user", "content": prompt}]
)

label = response.choices[0].message.content.strip()
print(f"Predicted label: {label}")

output

Predicted label: positive

Common variations

Use different models like gpt-4o-mini for faster, cheaper classification.
For async usage, use asyncio with the OpenAI client.
Use prompt templates to handle multiple texts or batch classification.
Incorporate streaming for real-time classification feedback (less common for classification).

python

import asyncio
from openai import OpenAI

async def classify_text_async(text, labels):
    client = OpenAI(api_key=os.environ["OPENAI_API_KEY"])
    prompt = (
        f"Classify the following text into one of these categories: {', '.join(labels)}.\n"
        f"Text: \"{text}\"\n"
        "Return only the label."
    )
    response = await client.chat.completions.create(
        model="gpt-4o-mini",
        messages=[{"role": "user", "content": prompt}]
    )
    return response.choices[0].message.content.strip()

async def main():
    label = await classify_text_async("The movie was boring and too long.", ["positive", "negative", "neutral"])
    print(f"Predicted label: {label}")

asyncio.run(main())

output

Predicted label: negative

Troubleshooting

If you get an authentication error, verify your OPENAI_API_KEY environment variable is set correctly.
If the model returns unexpected output, refine the prompt to be more explicit about returning only the label.
For rate limit errors, consider retrying with exponential backoff or upgrading your plan.

✅

Key Takeaways

Use chat.completions.create with a clear prompt for zero-shot classification.
Pass candidate labels explicitly in the prompt for best results.
Use gpt-4o for accuracy or gpt-4o-mini for cost-effective classification.
Always set your API key via environment variables to avoid authentication issues.
Refine prompts to ensure the model returns only the classification label.

Verified 2026-04 · gpt-4o, gpt-4o-mini

Verify ↗