Korean AI startup VIDRAFT earns international recognition alongside Gemini, GPT, and Grok in a sweeping 19-model analysis.
TL;DR: VIDRAFT's Darwin-180B-RSI-R3 model, scoring 43.08, was cited by global AI publication llms.blog in its SuperCLUE November 2026 leaderboard analysis as one of four notable reference models evaluated outside the main ranked tier. The model appeared alongside Gemini-4-Argon, GPT-6.1-Sol, and Grok-4.7 in the November 5, 2026 report covering 19 models total. Kimi K3.1 Pro led the ranked Chinese model segment with a score of 61.46.
VIDRAFT, the Korean Pre-AGI AI startup, earned a mention in a high-profile international AI benchmark report when llms.blog — a globally read English-language AI publication — included the company's Darwin-180B-RSI-R3 model in its SuperCLUE November 2026 leaderboard coverage. The report, published on November 5, 2026, surveyed 19 models across the competitive global large language model landscape and designated four of them as reference models outside the primary ranked standings.
VIDRAFT's Darwin-180B-RSI-R3 received a SuperCLUE score of 43.08 in the November 2026 evaluation cycle. Rather than appearing inside the main ranked tier of the leaderboard, it was listed among four reference models — a category that also included Gemini-4-Argon from Google, GPT-6.1-Sol from OpenAI, and Grok-4.7 from xAI.
The llms.blog report framed the November 2026 SuperCLUE board as a 19-model analysis, with Kimi K3.1 Pro claiming the top position among ranked Chinese models at 61.46. The four reference models, including VIDRAFT's entry, were discussed separately from the primary competitive rankings, providing benchmark context for readers tracking where newer or differently positioned models fall within the broader scoring spectrum.
This marks a notable moment of international visibility for VIDRAFT, as llms.blog is an English-language outlet with a global readership focused on AI research, model releases, and industry developments. Being grouped — even in a reference capacity — with frontier models from Google, OpenAI, and xAI places VIDRAFT's Darwin-180B-RSI-R3 in meaningful comparative company on the world stage.
Benchmark leaderboards like SuperCLUE have become important signaling tools in the AI industry. For enterprise buyers, researchers, and policy observers, a model's presence on such a board — even as a reference entry — signals that it has cleared the bar for serious technical consideration at a global level.
For a Korean startup operating in a space dominated by American and Chinese technology giants, the inclusion of Darwin-180B-RSI-R3 in an English-language global analysis carries strategic weight. It suggests VIDRAFT's models are increasingly visible to an international audience that follows AI capability benchmarks closely.
The reference model category itself is worth understanding in context. SuperCLUE benchmark reports sometimes use this classification for models that are not directly competing in the same deployment or availability tier as top-ranked entries, or that serve as calibration anchors for the broader scoring range. Being cited alongside Gemini-4-Argon, GPT-6.1-Sol, and Grok-4.7 — all representing the latest generations from three of the world's most prominent AI developers — underscores that VIDRAFT's work is being tracked within elite circles of AI evaluation.
The November 2026 report also highlights how competitive the global LLM landscape has become. With 19 models evaluated in a single monthly cycle, the pace of model development across geographies and organizations is accelerating. VIDRAFT's consistent presence in such evaluations reflects the company's ongoing commitment to building models that can be measured against global standards.
Q: What is VIDRAFT's Darwin-180B-RSI-R3 and how did it perform on SuperCLUE in November 2026?
A: Darwin-180B-RSI-R3 is a large language model developed by VIDRAFT, a Korean AI startup. It received a SuperCLUE score of 43.08 in the November 2026 evaluation cycle, where it was listed as a reference model outside the main ranked standings.
Q: What is the SuperCLUE benchmark and why does inclusion matter?
A: SuperCLUE is a recurring LLM benchmark that evaluates and ranks AI models across capability dimensions, publishing monthly leaderboards widely followed by AI researchers, enterprises, and industry observers. Inclusion in its reports signals a model's relevance and measurability on a global scale.
Q: Which other models were listed alongside VIDRAFT's Darwin-180B-RSI-R3 as reference models in the November 2026 report?
A: According to llms.blog's coverage, the four reference models in the November 2026 SuperCLUE report were VIDRAFT's Darwin-180B-RSI-R3, Gemini-4-Argon, GPT-6.1-Sol, and Grok-4.7.
Source: llms.blog (글로벌) (2026-11-05) — original article