AI News
LG Opens K-EXAONE 2.5 API Free of Charge
LG AI Research announced on November 9 that the API for its upgraded K-EXAONE 2.5 foundation model is now free for everyone. Developers, startups, companies and research institutes can get an API key through the EXAONE API Platform and use the model to build services and run research at no cost, starting today. The announcement was made at NVIDIA AI Day Seoul.
K-EXAONE 2.5 is a technical-validation model, built by further training K-EXAONE 2.0, which LG released in August as part of Korea's sovereign AI program. It is a text-only model in Korean and English, with 750 billion total parameters and 31 billion active, a 512,000-token context window, and an OpenAI-compatible chat completions endpoint (model id k-exaone_v2.5).
What LG claims it can do
LG says overall performance across 24 metrics rose more than 70% in eight weeks of further training. It also reports internal agentic scores of 59.2 on SWE-Bench Pro, 78.0 on BrowseComp, 74.6 on AutomationBench and 77.3 on Claw-Eval — roughly on par with DeepSeek-V4.1-Flash. These are LG's own figures, from its own testing, and they have not been independently reproduced.
The 70% figure needs the same care: LG has not published the benchmark suite or the baseline it measured against, so the number cannot be checked from the outside. Treat it as the company's claim about its own model, not a verified result.
What is actually free, and what is not
Free here means API access at no cost, not the model itself. The weights for K-EXAONE 2.5 have not been released; you can call the model through LG's platform, but you cannot download it and run it yourself. LG has also published no end date for the free period and no quota or rate limits, so the terms of "free" are open-ended for now.
That distinction matters for anyone planning to build on it. A free API with no published limits is useful for trying the model and for small projects; a product that depends on it needs a clear answer on what happens when the free period ends or when usage grows. LG has not given that answer yet.
No NVIDIA partnership — yet
LG said it is "considering" using NVIDIA Dynamo, the software for serving large models across GPUs, but there is no signed partnership with NVIDIA. The announcement was made at NVIDIA's event, which makes the connection look closer than it is: the free API is LG's own move, on LG's own platform.
How to try it
Get a key at docs.exaone.ai, the EXAONE API Platform. Because the endpoint follows OpenAI's chat completions format, existing code written for OpenAI-compatible APIs should work with only the base URL and model id changed. Start with small tests — and keep an eye on LG's pages for the quota and end-date details that are still missing.