A new AI leaderboard, whatmodel.sh, reveals that China's open-weight models are approximately seven months behind the leading US closed-weight models. The leaderboard compiles over 22,000 benchmark results from more than 1,000 sources, with a focus on text models. The data includes both independent evaluations and lab-reported scores, with all dataset and scoring code made publicly available. AI
IMPACT Provides a benchmark for evaluating the progress and competitive landscape of open-weight AI models compared to closed-weight models.
RANK_REASON The cluster presents data from an AI leaderboard that analyzes benchmark results, categorizing it as research. [lever_c_demoted from research: ic=1 ai=1.0]
AI-generated summary · Google Gemini · from 1 sources. How we write summaries →