LG opens free API access to K-EXAONE 2.5, its 750-billion-parameter AI model

LG AI Research announced on Monday, November 9, 2026, free API access to K-EXAONE 2.5, its proprietary artificial intelligence model. The announcement was made during NVIDIA AI Day Seoul 2026 at COEX in Seoul, South Korea. Geekiki has already covered the event, including the announcements from NVIDIA, SK Telecom and Rebellions on sovereign AI infrastructure.

What is K-EXAONE 2.5

K-EXAONE 2.5 is a technology-validation model that received additional post-training on top of K-EXAONE 2.0, released in August 2026. According to LG, K-EXAONE 2.5 has 750 billion parameters, with 31 billion active per token, works with text only in Korean and English, and supports a 512K-token context window.

Agentic AI benchmark results

In internal evaluations, LG says the overall performance of K-EXAONE 2.5 improved by more than 70% over the August model, across 24 combined metrics. In agentic AI tests, the company reported 74.6 on AutomationBench-Openset, 59.2 on SWE-Bench Pro, 78.0 on BrowseComp and 77.3 on Claw-Eval — results that, according to LG, put K-EXAONE 2.5 on a competitive footing with models such as DeepSeek V4.1 Flash and GLM 5.3. These figures are LG's internal measurements, not yet independently verified, and should be read as the company's claims rather than third-party-confirmed facts.

How to use the free K-EXAONE 2.5 API

Access to K-EXAONE 2.5 is provided through the EXAONE API Platform (docs.exaone.ai): developers, startups, companies and research institutions can obtain an API key and use K-EXAONE 2.5 at no cost for service development and research. The interface is compatible with OpenAI's API, using the model identifier k-exaone_v2.5. LG has not said how long free access will last.

Besides K-EXAONE 2.5, the platform also lists EXAONE Tabular, a smaller model for predictions on structured data, according to the platform's documentation.

What comes next for LG

The company is also studying NVIDIA's Dynamo technology to improve the model's inference efficiency. According to LG AI Research, the goal is to complete a frontier-level EXAONE model by January 2027, using the technologies validated in K-EXAONE 2.5.

Image credit: image released by LG AI Research, via DigitalToday.

Sources: LG AI Research — K-EXAONE 2.5 API Available for Free!; DigitalToday — LG AI Research opens K-EXAONE 2.5 API for free; Unite.AI — LG AI Research Opens Free API Access to Its K-EXAONE 2.5 Model; Aju Press — LG AI Research Opens Free Access to 'K-Exaone 2.5' API.

Transparency: this article was produced with the assistance of artificial intelligence, cross-checking public posts and internet sources. We recommend consulting the original sources to confirm the information.

By Muse