How a 180B Open Model Outscored a 2.8T Model in Math and Science Reasoning: Strength Merging Plus Model-Level RSI
TL;DR Darwin-180B-RSI , an open-weight model from Korea's VIDRAFT, scored 73.68 in mathematical and scientific reasoning on SuperCLUE's November 2026 evaluation. That is first among open-weight models and second overall, directly behind OpenAI's GPT-6.1-Sol (76.32). The largest model in the same evaluation, Moonshot AI's Kimi-K3.1-Pro at 2.8 trillion parameters, scored 63.16. Darwin has 180…
In the November 2026 SuperCLUE evaluation, an open-weight model called Darwin-180B-RSI from Korea's VIDRAFT company achieved a score of 73.68 in mathematical and scientific reasoning. This is the highest score among open-weight models and the second overall, just behind OpenAI's GPT-6.1-Sol with a score of 76.32. The largest model in the evaluation, Moonshot AI's Kimi-K3.1-Pro with 2.8 trillion parameters, scored 63.16, indicating that a 180 billion parameter model can outperform a much larger one.
This post explores the two methods behind Darwin's success and the implications of the size difference for serving costs.
Written by urgent.news from Dev.to's reporting — not their text. Machine-written — may contain errors; check the original before relying on it.