← 全部模型 Gemini · 近 7 天Gemini 2.5 Pro0–100 分，越高代表社区使用体验越正面，不代表能力测试成绩。更新于 2026-09-05 14:21 UTC · 2026-08-29 – 2026-09-05 UTC社区口碑分评价积累中3 条评论 · 近 7 天 社区评论精选近 7 天的不同意见，摘录条数不代表真实好评比例。全部摘录正面负面2 条精选摘录 负面通用文本2026-09-03 “我记得Gemini 2.5 pro的输出有令人抓狂的不一致性。” Redditen机器翻译 展开原文I remember gemini 2.5 pro having crazy inconsistent outputs.正面推理2026-09-02 “如果你知道怎么用的话，Gemini 2.5 pro比你想象的要好得多，它在实际的逻辑/符号推理方面比大多数新模型都要强得多” Redditen机器翻译 展开原文Gemini 2.5 pro is way better than you'd think if you knew how to use it, it was far better at actual logical/symbolic reasoning than most of the new models正面编程2026-08-30 “直到Gemini 2.5 pro，模型才能输出始终可执行的代码，没有运行时错误，零样本” Redditen机器翻译 展开原文It wasn't until Gemini 2.5 pro where models were putting out consistently executable code, no runtime errors, zero shot 近 7 天暂时没有符合此筛选条件的精选评论。查看更多摘录
负面通用文本2026-09-03 “我记得Gemini 2.5 pro的输出有令人抓狂的不一致性。” Redditen机器翻译 展开原文I remember gemini 2.5 pro having crazy inconsistent outputs.
正面推理2026-09-02 “如果你知道怎么用的话，Gemini 2.5 pro比你想象的要好得多，它在实际的逻辑/符号推理方面比大多数新模型都要强得多” Redditen机器翻译 展开原文Gemini 2.5 pro is way better than you'd think if you knew how to use it, it was far better at actual logical/symbolic reasoning than most of the new models
正面编程2026-08-30 “直到Gemini 2.5 pro，模型才能输出始终可执行的代码，没有运行时错误，零样本” Redditen机器翻译 展开原文It wasn't until Gemini 2.5 pro where models were putting out consistently executable code, no runtime errors, zero shot
社区评论
精选近 7 天的不同意见，摘录条数不代表真实好评比例。
2 条精选摘录
展开原文
I remember gemini 2.5 pro having crazy inconsistent outputs.
展开原文
Gemini 2.5 pro is way better than you'd think if you knew how to use it, it was far better at actual logical/symbolic reasoning than most of the new models
展开原文
It wasn't until Gemini 2.5 pro where models were putting out consistently executable code, no runtime errors, zero shot
近 7 天暂时没有符合此筛选条件的精选评论。