Seoul Shinmun Reports Darwin First Among Open-Weight Models in SuperCLUE Math and Science Reasoning

2026-11-13 · VIDRAFT AI Media

aillmbenchmarknews

Seoul Shinmun Reports Darwin First Among Open-Weight Models in SuperCLUE Math and Science Reasoning

TL;DR

On November 13, 2026, Seoul Shinmun, a national daily newspaper in Korea, reported that VIDRAFT's reasoning model Darwin-180B-RSI recorded the highest score among open-weight models in the mathematical and scientific reasoning track of SuperCLUE's November 2026 evaluation. Across all 19 models, including closed ones, Darwin placed second, directly behind OpenAI's GPT-6.1-Sol.

SuperCLUE November 2026 math and science reasoning results: all 16 ranked models

What Seoul Shinmun reported

Point What the article says
Result First among open-weight models, second of 19 overall in math and science reasoning
Models placed below Google Gemini, xAI Grok, DeepSeek, Moonshot AI Kimi
Overall index Fifth among open-weight models, ninth of 19 across the five tracks
Size 180 billion parameters, scoring above a 2.8 trillion parameter model
Hugging Face records GPQA 94.44, AIME 2026 and HMMT 100, results on LEXam and ExtractBench
Method Model synthesis from the strengths of parent models, then model-level recursive self-improvement
Safety ViKernel, rules in the model weights that judge actions and let a person stop the model
Hardware VKORE, training and running models on Korean NPUs such as Rebellions ATOM-Max
On-device The POCKET series runs on a laptop

The article closes with CEO Kim Minsik's comment that the result carries weight because an outside evaluator graded the model directly.

Why this coverage matters

Seoul Shinmun is a general-interest national daily. Its coverage follows reports by Chosun Biz, Economist Korea, Digital Daily and AI Times on November 6, and takes the result from the technology press to a general readership. The article also describes the full technology stack in one place: how the model is built, how it is kept safe, and where it runs.

The numbers

Mathematical and scientific reasoning, SuperCLUE November 2026:

Rank Model Developer Open weights Score
1 GPT-6.1-Sol OpenAI No 76.32
2 Darwin-180B-RSI-R3 VIDRAFT Yes 73.68
3 Gemini-4-Argon Google No 71.05
3 DeepSeek-V4.1-Flash DeepSeek Yes 71.05
5 Grok-4.7 xAI No 68.42

SuperCLUE grades each model itself on a private test set that changes every month. Developers do not submit scores.

  • Seoul Shinmun article (Korean): https://www.seoul.co.kr/news/economy/2026/11/13/20261113500198
  • SuperCLUE board: https://www.superclueai.com/homepage
  • Model: https://huggingface.co/FINAL-Bench/Darwin-180B-RSI-R3
  • Laptop edition: https://huggingface.co/FINAL-Bench/POCKET-Darwin-180B-GGUF
  • VIDRAFT press list: https://vidraft.net