exaone

An unofficial Python client for the EXAONE API Platform, with sync and async access to its endpoints.

Caution

This project is not affiliated with or endorsed by LG AI Research. EXAONE API use is subject to the EXAONE API Platform Terms of Service. The terms limit API and model use to non-commercial purposes. They prohibit commercial use of the model or its outputs, and integration into a product or service that lets third parties call the model or submit inputs. Check the terms before using this client.

Install

uv add git+https://github.com/RektPunk/exaone.git
# or
pip install git+https://github.com/RektPunk/exaone.git

Set your API key

The client reads EXAONE_API_KEY from the process environment. It does not load .env files itself.

# macOS / Linux
export EXAONE_API_KEY="your-api-key"

# Windows PowerShell
$env:EXAONE_API_KEY = "your-api-key"

Chat

import asyncio

from exaone import AsyncExaone, ChatCompletionRequest, ChatMessageRequest, Exaone

request = ChatCompletionRequest(
    messages=[ChatMessageRequest(role="user", content="Hi!")],
    reasoning_effort="low",
)


def sync_examples():
    with Exaone() as client:
        # Synchronous chat
        response = client.chat.create(request)
        print(response.choices[0].message.content)

        # Synchronous streaming
        for chunk in client.chat.stream(request):
            if chunk.choices:
                content = chunk.choices[0].delta.content
                if content:
                    print(content, end="", flush=True)
        print()


async def async_examples():
    async with AsyncExaone() as client:
        # Asynchronous chat
        response = await client.chat.create(request)
        print(response.choices[0].message.content)

        # Asynchronous streaming
        async for chunk in client.chat.stream(request):
            if chunk.choices:
                content = chunk.choices[0].delta.content
                if content:
                    print(content, end="", flush=True)
        print()


sync_examples()
asyncio.run(async_examples())

Tabular prediction

Pass a training CSV with the target column and a test CSV containing only the feature columns for prediction.

from exaone import ClassificationRequest, Exaone, RegressionRequest

classification_request = ClassificationRequest(
    target_columns=["churn"],
    output_type="probability",
)
regression_request = RegressionRequest(
    target_columns=["price"],
    output_type="point",
)

with Exaone() as client:
    # Classification
    classification = client.tabular.classify(
        "classification_train.csv",
        "classification_test.csv",
        classification_request,
    )
    print("Classes:", classification.data.classes)
    print("Probabilities:", classification.data.probabilities)

    # Regression
    regression = client.tabular.regress(
        "regression_train.csv",
        "regression_test.csv",
        regression_request,
    )
    print("Predictions:", regression.data.predictions)