Running 111 111 Open-LLM performances are plateauing, letβs make the leaderboard steep again π Update leaderboard for fair model evaluation