Fetching from the wire…
01
02
03
04
05
06
07
08
Models2026-08-02 · source-backed
Released July 31 under South Korea's Sovereign AI Foundation Model Project: hybrid-attention MoE, 750B total with ~37B active per token, 256 experts with 8 selected, a 262,144-token context window, 10-language coverage. Over 3x the size of the first K-EXAONE (236B/23B), lifting the average benchmark score from 63.3 to 70.1, with 83.5 MMLU-Pro and 92.3 AIME 2026. Apache 2.0 makes it one of the largest permissively-licensed open-weight models available. Sovereign AI programs producing genuinely competitive open weights is a 2026 development I did not expect at the start of the year.
Each link below shares sources, entities, or timing with this story.
Shared entities / Shared topic / Earlier coverage
Both cover AIME, Apache, MoE; overlapping topics (active, aime, apache, model); earlier AIME coverage from 2026-02-20.
Shared entities / Shared topic / Earlier coverage / Tension
Both cover Apache, MMLU, MoE; overlapping topics (apache, context); earlier Apache coverage from 2026-03-15.
Both cover AIME, MoE; overlapping topics (aime, benchmark, model, token); earlier AIME coverage from 2026-06-07.
Shared entities / Shared topic / Earlier coverage
Both cover AIME, MMLU, MoE; overlapping topics (benchmark, model); earlier AIME coverage from 2026-06-14.
Both cover AIME, Apache, MoE; overlapping topics (apache, model); earlier AIME coverage from 2026-04-13.
Shared entities / Shared topic / Earlier coverage / Tension
Both cover Apache, MMLU; overlapping topics (apache, model, token); earlier Apache coverage from 2026-04-01.
Shared entities / Earlier coverage / Tension
Both cover Apache, MMLU, MoE; earlier Apache coverage from 2026-06-27; pushes against this story (vs).
Shared entities / Shared topic / Earlier coverage
Both cover AIME, Over; overlapping topics (active, benchmark, context, model); earlier AIME coverage from 2026-04-08.