Korea enters the open-weights heavyweight ring: LG's 750B K-EXAONE 2.0 ships under Apache 2.0
LG AI Research released K-EXAONE 2.0, the largest AI model ever built in South Korea: a 750-billion-parameter mixture-of-experts (37B active), triple the size of its predecessor, with a 262K context window and support for 10 languages. It averaged 70.1 across 24 benchmarks — up from 63.3 for version 1.0 — and beat Zhipu's GLM-5.1 on long-context tests. Unusually for a national flagship model, it ships on Hugging Face under Apache 2.0, allowing unrestricted commercial use.
Why it mattersThe open-weights race stopped being a US-versus-China story this week: Korea's biggest-ever model arrived with a fully permissive license on the same day Huawei and MiniMax made their own open releases. For companies building on open models, the menu of frontier-scale options — and the leverage against closed-model pricing — keeps growing.
Read in the app — free, in 9 languages
Related stories
DeepSeek's cheap workhorse can now see — and on agent tasks that need eyes it says it is close to Anthropic's best2026-08-21Show the robot once — three to twelve seconds — and it gets the job right 59 times out of 100 with no training at all
2026-08-21The model invents its own tasks, builds the rig to test them, then trains on the results — and DeepReinforce gave the weights away
2026-08-20A free model that fits on one graphics card scores 52 — then spends 22,276 thinking tokens drawing a picture
2026-08-18Zhipu did not build a new model. It kept training the old one — and says coding got 50 percent better.
2026-08-17