Rohan Paul@rohanpaul_ai · XAI 摘要
GPT-5.5 Pro在Epoch AI的ECI指数中获得159分，创下该综合能力评分的新纪录，表明其在多项测试中的整体模型能力目前领先。ECI指数旨在通过整合多个基准测试来评估模型的广泛推理能力，避免单一测试易被针对性优化的问题。此外，GPT-5.5 Pro与GPT-5.5共同解决了两个此前未有模型能处理的Tier 4级别难题。
GPT-5.5 Pro just posted the highest Epoch AI's ECI score yet at 159， which means it now leads their combined measure of general model capability across many separate tests.
ECI matters because single benchmarks are easy to game or overfit， while a pooled index tries to measure whether a model is getting broadly better at many kinds of reasoning at once rather than just spiking on one exam.
GPT-5.5 Pro and GPT-5.5 also solved 2 Tier 4 problems that no earlier model had solved.