LG enters global AI race with hybrid Exaone 4.0 model
LG AI Research, the artificial intelligence arm of LG Group, on Tuesday unveiled Exaone 4.0, Korea's first "hybrid" AI model that combines the language fluency of large language models with the advanced problem-solving abilities of reasoning AI.
This breakthrough represents a major leap in Korea's ambition to lead in next-generation AI, as the model can both understand human language and logically reason through complex tasks.
'We will continue to advance research and development so that Exaone can become Korea's leading frontier AI model and prove its competitiveness in the global market,' said LG AI Research's Exaone lab leader Lee Jin-sik.
Globally, only a few companies have announced similar hybrid AI architectures, including US-based Anthropic with its Claude model and China's Alibaba with Qwen. OpenAI is reportedly working on GPT-5, which is expected to adopt a hybrid structure as well.
In benchmark evaluations across a variety of domains — including knowledge comprehension, problem-solving, coding, scientific reasoning and mathematics — Exaone 4.0 outperformed leading open-weight models from the US, China and France, establishing itself as one of the world's most capable AI models.
LG introduced two versions of the model: a 32B expert model with 32 billion parameters and a lighter 1.2B on-device model with 1.2 billion parameters.
The expert model successfully passed written exams for six national professional licenses, including those for physicians, dentists, Korean medicine doctors, customs brokers, appraisers and insurance adjusters — underscoring its domain-level expertise.
The on-device model, designed for practical use in consumer electronics such as home appliances, smartphones, automotive infotainment systems and robots, runs independently on devices without connecting to external servers. It ensures faster processing, enhanced privacy, and robust security, which are key advantages in an increasingly AI-integrated world.
Despite being half the size of its predecessor, the Exaone 3.5 2.4B model released in December, the new on-device model surpasses GPT-4o mini from OpenAI in specialized evaluations across mathematics, coding, and science, making it the most powerful in its weight class globally.
To accelerate open research and innovation, LG has released Exaone 4.0 as an open-weight model on Hugging Face, the leading global platform for open-source AI. While open-weight models do not disclose architectural blueprints or training data, they make the trained weights publicly accessible, allowing developers to fine-tune and redistribute the models.
Notable peers in the open-weight category include Google's Gemma, Meta's LLaMA, Microsoft's Phi, Alibaba's Qwen and Mistral AI's Mistral.
Additionally, LG has partnered with Friendly AI, an official Hugging Face distribution partner, to launch a commercial API service for Exaone 4.0. It allows developers and enterprises to easily deploy the model without requiring high-end GPUs, thereby broadening accessibility across industries.
In his New Year's address, LG Group Chairman Koo Kwang-mo underscored the group's long-term vision for artificial intelligence, calling it a core driver of future innovation.
'LG has grown by constantly pioneering uncharted territory and creating new value,' he said. 'Now, with AI, we aim to reshape everyday life by making advanced technologies more accessible and meaningful, helping people reclaim their time for what truly matters.'
He emphasized that LG's AI efforts are not just about technological leadership, but about enabling a smarter, more human-centered lifestyle through seamless AI integration across products and services.
Meanwhile, LG hosted 'Exaone Partners Day' on Tuesday, bringing together 22 domestic partner companies to discuss strategies for expanding the Exaone ecosystem. The group will further highlight its AI ambitions at the 'LG AI Talk Concert 2025' in Seoul on July 22, where it plans to unveil its latest research breakthroughs and present a roadmap for future AI innovation.
Hashtags
Try Our AI Features
Explore what Daily8 AI can do for you:
Comments
No comments yet...
Related Articles
![[Yoo Choon-sik] Factories of the future: Lee's vision beyond AI](/_next/image?url=https%3A%2F%2Fwimg.heraldcorp.com%2Fnews%2Fcms%2F2025%2F07%2F20%2Fnews-p.v1.20250720.68b137f432784c0fb2e58ec66b79a701_T1.jpg&w=3840&q=100)
![[Yoo Choon-sik] Factories of the future: Lee's vision beyond AI](/_next/image?url=https%3A%2F%2Fall-logos-bucket.s3.amazonaws.com%2Fkoreaherald.com.png&w=48&q=75){kind=small}
Korea Herald
an hour ago
- Korea Herald
[Yoo Choon-sik] Factories of the future: Lee's vision beyond AI
South Korea has long prided itself on being one of the few nations capable of fully leveraging the potential of artificial intelligence innovation. This national confidence is rooted in the country's remarkable transformation from economic crisis to technological leadership ― a journey that began in earnest during the aftermath of the Asian financial crisis in the late 1990s. That legacy, however, now faces a new test. Recent independent studies have raised alarms that South Korea may be slipping behind in the global race to lead the AI revolution. One such report, published by the Australia-based technology firm Cloudscene, placed South Korea at 30th worldwide in terms of the number of AI-usable data centers. Shockingly, this ranking puts the country behind many others that it had once viewed as trailing in the domain of technological innovation. This surprising placement has prompted concern among policymakers and industry experts alike. Many attribute South Korea's lagging performance to policy missteps under the previous Yoon Suk Yeol administration, combined with a persistent failure to overhaul outdated administrative regulations. While the previous government emphasized more efficient use of research funds, it overlooked the importance of improving the business environment for tech infrastructure ― particularly the construction and operation of data centers by both domestic and international firms. As recently as early 2024, skepticism remained widespread ― even within government ranks ― about the notion that the global competition to develop stronger foundational AI models would continue to escalate. Many also questioned whether it was truly necessary for South Korea to dramatically scale up its computing power by securing more high-performance graphics processing units, especially given the massive costs involved. These doubts were swiftly proven wrong. As AI development intensified globally, the price of GPUs surged amid fierce competition among the world's tech giants. Companies like OpenAI, Google, Microsoft, Meta, Anthropic and Alibaba began pouring unprecedented investments into expanding their data center capacity and equipping them with the most advanced chips on the market. Despite this global momentum, debate persisted in South Korea. Some policymakers argued that the country could not feasibly compete with the world's largest tech conglomerates in developing foundational AI models. Instead, they suggested focusing on building secondary applications ― tools and services that piggyback on the capabilities of existing models ― once the foundational development race had settled. While this pragmatic approach had its merits, it overlooked a critical opportunity. Even if South Korea did not lead in model development, it could still serve as a vital AI hub by attracting global firms to build their data centers within its borders. The reality, however, is stark. Cloudscene's report found that Indonesia and Malaysia had 84 and 62 AI-ready data centers, respectively, compared to South Korea's 43. This disparity is especially troubling given South Korea's stronger standing in broader indices of technological and economic development. President Lee's ambitious AI policy Fortunately, a new sense of urgency has emerged under the leadership of President Lee Jae Myung. Unlike his predecessor, Lee has recognized the transformative power of AI and the importance of establishing a robust infrastructure to support it. At a groundbreaking ceremony for a new data center in Ulsan ― a joint venture between SK Group and Amazon Web Services ― Lee drew an evocative parallel between the digital infrastructure of the future and the Gyeongbu Expressway of the past. That expressway, constructed in the 1960s against widespread skepticism, played a critical role in catalyzing South Korea's industrial rise. Although he has not announced a specific target for the number of data centers to be built during his five-year term, Lee did pledge during his election campaign to procure 50,000 high-end GPUs. This would represent a massive leap forward, especially considering that only several thousand GPUs are believed to be currently in operation across the country. Equally significant is Lee's vision to decentralize this new digital infrastructure. Rather than concentrating all development in the capital region, his administration is seeking to distribute data centers across a wide range of provinces. This strategy not only addresses the current shortfall in AI computing capacity, but also tackles one of South Korea's most persistent social challenges: The overconcentration of population and economic activity in the Seoul metropolitan area. As of 2024, over 51 percent of the population lived in Seoul, Incheon and Gyeonggi Province ― an area that comprises just 12 percent of the nation's landmass. Modern data centers have evolved far beyond their original function as basic storage facilities. Today, they serve as industrial-scale computational hubs essential for training, fine-tuning, and deploying advanced AI models. Establishing such facilities outside the capital region could help balance national development by generating jobs both directly and indirectly related to the AI industry. According to the Korea Data Center Council, private-sector revenue from data centers is projected to grow by an average of 13 percent annually over the next four years ― even before accounting for the anticipated expansion under Lee's administration. However, these gains will only materialize if structural challenges are addressed. Chief among them is the country's aging and inefficient power grid system, which remains a bottleneck for large-scale AI infrastructure projects. In response, Lee has pledged to implement regulatory reforms and offer incentives to companies that locate new data centers near renewable energy sources in the provinces. Still, the key lies in translating these promises effectively into action. Reforming entrenched systems takes time, political capital, and cross-sector cooperation. Yet the urgency is undeniable. South Korea cannot afford to wait for the perfect moment. The global AI landscape is evolving rapidly. For South Korea, the stakes are more than economic. They touch on national identity and the country's ability to shape its future in an increasingly AI-driven world. Whether the country emerges as a leader or falls behind may depend on the decisions made in the next few years. The groundwork is being laid ― not just for faster computation, but for a new era of growth, equity and innovation.
Korea Herald
17 hours ago
- Korea Herald
LG Uplus debuts financial AI model on AWS Bedrock
LG Uplus, one of the top three telecommunications providers in South Korea, said Sunday that it has launched its specialized small language model, ixi-Gen, on the Amazon Web Services Bedrock marketplace ― becoming the first Korean telecom company to bring an internally developed artificial intelligence language model to the global stage. Built on LG's large language model Exaone, ixi-Gen has been optimized for financial applications through targeted learning on finance-specific data sets. According to LG Uplus, the model delivers up to 31 percent improved performance in financial tasks compared to general-purpose AI models, while maintaining high levels of accuracy, safety and content filtering. AWS Bedrock is a fully managed AI service that enables developers to build applications using foundation models from leading providers without managing infrastructure. Its marketplace functions as a central hub for organizations to explore, compare and integrate AI models ― offering over 100 models globally from major players such as Amazon, Meta, Mistral and Cohere. To be listed on the Bedrock marketplace, AI models must pass rigorous security and performance evaluations. LG Uplus emphasized that ixi-Gen's successful listing reflects both its technical sophistication and trustworthiness ― qualities increasingly sought after by enterprises adopting AI. The mobile carrier plans to expand the capabilities of ixi-Gen beyond the financial and telecommunications sectors into other industries where domain-specific AI solutions are in demand.
Korea Herald
17 hours ago
- Korea Herald
Tech CEO quits after concert incident
The IT company CEO captured in a widely circulated video showing him embracing an employee at a Coldplay concert has resigned. Andy Byron resigned from his job as CEO of Cincinnati-based Astronomer, according to a statement posted on LinkedIn by the company Saturday. 'Astronomer is committed to the values and culture that have guided us since our founding. Our leaders are expected to set the standard in both conduct and accountability, and recently, that standard was not met,' the company said in its post on LinkedIn. The move comes a day after the company said that Byron had been placed on leave and the board of directors had launched a formal investigation into the jumbotron incident, which went viral. A company spokesman later confirmed in a statement to AP that it was Byron and Astronomer chief people officer Kristin Cabot in the video. The short video clip shows Byron and Cabot as captured on the jumbotron at Gillette Stadium in Foxborough, Massachusetts, during a Coldplay concert on Wednesday. Lead singer Chris Martin asked the cameras to scan the crowd for his 'Jumbotron Song,' when he sings a few lines about the people the camera lands on. 'Either they're having an affair or they're just very shy,' he joked. Internet sleuths identified the man as the chief executive officer of a US-based company and the woman as its chief people officer. Pete DeJoy, Astronomer's cofounder and chief product officer, has been tapped as interim CEO while the company conducts a search for Byron's successor. It's easy to miss, but most concert venues have signs informing the audience that they could be filmed during the event. Look for them on the walls when you arrive and around the bar areas or toilets. It's common practice especially when bands like to use performances for music videos or concert films. The venue in this case, Gillette Stadium in Foxborough, also has a privacy policy online which states, 'When you visit our location or attend or participate in an event at our location, we may capture your image, voice and/or likeness, including through the use of CCTV cameras and/or when we film or photograph you in a public location.' 'They probably would have got away with it if they hadn't reacted,' said Alison Taylor, a clinical associate professor at New York University's Stern School of Business. And by the time the alleged identities emerged on social media, it hit a classic nerve around 'leaders acting like the rules don't apply to them,' she added. Still, Taylor and others stress how quickly such a video lead to an internet search to find the people involved — and note that it's important to remember that such 'doxing' isn't just reserved for famous people. Beyond someone simply spotting a familiar face and spreading the word, technological advances, such as the rising adoption of artificial intelligence, has made it easier and faster overall to find just about anyone in a viral video today. 'It's a little bit unsettling how easily we can be identified with biometrics, how our faces are online, how social media can track us — and how the internet has gone from being a place of interaction, to a gigantic surveillance system,' said Mary Angela Bock, an associate professor in the University of Texas at Austin's School of Journalism and Media. 'When you think about it, we are being surveilled by our social media. They're tracking us in exchange for entertaining us.' (AP)
