Doubao's Logic Performance Falters After Update, Gemini Emerges Superior

Users conducted a hands-on performance test on Doubao’s upgraded AI model, evaluating its processing capabilities with a two-image logic puzzle. The results showed that in Superpower Mode, Doubao focused solely on searching without effectively applying the rules; in Thinking Mode, it completely ignored the rule instructions, leading to a failed solution. In contrast, the Gemini 3 Pro Preview model successfully solved the problem after two corrections. This comparison highlights the significant differences in logical reasoning abilities between AI models. Doubao’s performance after the upgrade fell short of expectations, while Gemini demonstrated stronger problem-solving skills. For users focused on AI technology, such practical performance comparisons offer valuable insights, helping to understand the strengths and weaknesses of each model and driving continuous improvement and optimization of AI technology.

Original Link:Linux.do

C code80.ai · AI 编码 API 聚合 Claude / GPT 多模型统一接入,稳定不限速,按量计费,几行配置接入 Claude Code。 了解一下 ›

抢沙发

评论前必须登录!

立即登录   注册