This page may contain third-party content, which is provided for information purposes only (not representations/warranties) and should not be considered as an endorsement of its views by Gate, nor as financial or professional advice. See Disclaimer for details.
OpenAI çalışanı Grok3'ün Benchmark test sonuçlarının yanıltıcı olduğunu açıkça belirtti
金十数据2月23日讯,近日,OpenAI的一名员工公开指责埃隆・马斯克旗下的xAI公司,称其发布的最新AI模型Grok3的Benchmark测试结果具有误导性。对此,xAI的联合创始人伊戈尔・巴布什金(Igor Babushkin)则坚称公司并无不当。xAI的图表显示,Grok3的两个版本——Grok3 Reasoning Beta和Grok3 mini Reasoning——在AIME 2025上的表现超过了OpenAI当前最强的可用模型o3-mini-high。然而,OpenAI的员工很快在X平台上指出,xAI的图表并未包含o3-mini-high在“cons@64”条件下的AIME 2025得分。巴布什金在X平台上辩称,OpenAI过去也曾发布过类似的误导性Benchmark测试图表。尽管这些图表是用于比较其自身模型的表现。