EXAONE 4.0 - Hybrid Reasoning Model introduced by LG

What is EXAONE 4.0

EXAONE 4.0 is a hybrid reasoning grand model from LG AI Research in Korea, blending general-purpose natural language processing and advanced reasoning capabilities. The model supports Korean, English, and Spanish, and is divided into a 32B professional version and a 1.2B end-side version. The professional version is suitable for highly specialized fields such as law, accounting, medicine, etc. The end-side version is small in size and high in performance, and is suitable for local operation scenarios with high privacy and security requirements. The model excels in complex task processing, such as scoring 81.8 in MMLU-Pro and 85.3 in AIME 2025, demonstrating strong complex task processing capability, and supporting function calls and MCP interfaces, which facilitates integration with other systems.EXAONE 4.0 is widely used in the fields of intelligent customer service, educational assistance, medical and healthcare, programming assistance and corporate office, etc., to Enhance efficiency and user experience.

Main features of EXAONE 4.0

• complex reasoning: Ability to efficiently solve complex tasks in science, math, and programming, supporting step-by-step thinking and logical reasoning.
• Multi-language support: Coverage of Korean, English and Spanish to enhance the model's global applicability.
• System Integration Capability: Supports function calls and MCP interfaces for easy interaction with other systems and tools to automate complex tasks.
• Dual version design::
  • Professional Edition (32B): Training based on knowledge in highly specialized fields such as law, accounting, medicine, etc., applicable to highly specialized scenarios.
  • End-side version (1.2B): Small size and support for local operation, suitable for scenarios with high privacy and security requirements.
• Education and Business Applications: Supports free use by educational institutions, capable of generating practice questions, correcting assignments, and providing personalized learning suggestions. Provide commercial API services to help enterprises quickly integrate and improve office efficiency.

EXAONE 4.0 official website address

• Project website:: https://www.lgresearch.ai/blog/view?seq=575
• HuggingFace Model Library::https://huggingface.co/collections/LGAI-EXAONE/exaone-40-686b2e0069800c835ed48375
• Technical Papers:: https://www.lgresearch.ai/data/cdn/upload/EXAONE_4_0.pdf

How to use EXAONE 4.0

• Hugging Face Platform Usage::
  • Choosing the right model version: Select the model version as required.
  • Download models for local deployment: Download the appropriate model files to load and use with Hugging Face's Python libraries (e.g. transformers). Example:

from transformers import AutoModelForCausalLM, AutoTokenizer

model_name = "LGAI-EXAONE/EXAONE-4.0-1.2B"
tokenizer = AutoTokenizer.from_pretrained(model_name)
model = AutoModelForCausalLM.from_pretrained(model_name)

input_text = "你的输入文本"
inputs = tokenizer(input_text, return_tensors="pt")
outputs = model.generate(**inputs)
print(tokenizer.decode(outputs[0], skip_special_tokens=True))

• API call::
  • Register for an account: Visit LG AI Research's official website or the relevant API platform to register an account and obtain an API key.
  • Read the document: Consult the technical documentation for EXAONE 4.0 to understand the API call methods, parameter settings, and return value formats.
  • Write code: Write code using programming languages such as Python, Java, etc. to call the EXAONE 4.0 API interface based on HTTP requests. Example:

import requests

url = "https://api.lgresearch.ai/exaone-4.0"  # 示例API地址
headers = {
    "Authorization": "Bearer YOUR_API_KEY",
    "Content-Type": "application/json"
}
data = {
    "prompt": "你的问题或任务描述",
    "language": "en"  # 语言选择，如"en"、"ko"、"es"
}

response = requests.post(url, headers=headers, json=data)
result = response.json()
print(result)

• • Processing return results: Based on the data format returned by the API, parse the results and apply them to actual business scenarios.

Core Benefits of EXAONE 4.0

• Powerful reasoning: Efficiently solves complex problems and supports step-by-step thinking and logical reasoning.
• Multi-language support: Covers Korean, English and Spanish for a wide range of applications.
• Flexible versioning: Professional version for highly specialized areas, end-side version for privacy and security demanding scenarios.
• Excellent system integration capability: Supports function calls and MCP interfaces for easy interaction with other systems and tools.
• Outstanding performance: Outperforms in demanding international benchmarks such as MMLU-Pro and AIME 2025.
• Wide range of application scenarios: Apply to a variety of fields such as intelligent customer service, educational assistance, healthcare, programming assistance and corporate office.

Who is EXAONE 4.0 for?

• Practitioners in specialized fields: lawyers, accountants, doctors, etc., to efficiently handle complex professional tasks.
• Educators and students: Educational institutions generate practice questions, correct homework, and provide personalized learning suggestions for students to help teaching and learning.
• corporations: Enterprises quickly integrate models to improve office efficiency and optimize workflow.
• customer service personnel: Intelligent customer service scenarios to quickly respond to customer inquiries, accurately answer complex questions, and improve customer satisfaction.
• programmer: Helps programmers generate code snippets, debug code, provide programming advice, and improve development efficiency.