meo-benchmark: Model Intelligence Leaderboard (dataset)

The meo-benchmark leaderboard dataset: first-party meo scores, the Effective Value (V) metric, efficiency, threat-susceptibility robustness, and license-clean aggregated third-party benchmarks, per model. Mirrors the Hugging Face and Kaggle releases. The private holdout (questions, answers, canaries) is never included. License CC-BY-4.0; Artificial-Analysis values excluded pending a commercial license.meo-benchmark排行榜数据集：按模型维度整理的自研meo评分、有效价值（Effective Value，V）指标、效率、抗威胁敏感性鲁棒性，以及经许可合规的第三方基准测试聚合结果。该数据集与Hugging Face及Kaggle平台发布的版本保持一致，且从未纳入私有预留集（含测试题目、参考答案、金丝雀样例）。本数据集授权协议为CC-BY-4.0；Artificial-Analysis相关数值暂未纳入，待获取商业授权后补充。