← 全部模型 Qwen · 近 7 天Qwen3.5-9B0–100 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58.7/10010 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1 条精选摘录 负面编程2026-09-05 “Qwen会提供备份和打印输出，但总是有1-3个错误” Redditen机器翻译 展开原文Qwen will offer backups and printouts but always has 1-3 errors正面速度与延迟2026-09-05 “我可以和Qwen3.5-9b以及更小的模型进行有趣的对话” Redditen机器翻译 展开原文I can have fun chats with Qwen3.5-9b and anything smaller正面速度与延迟2026-09-03 “我用qwen3.5-9b配合dflash大概能跑到100-200 t/s。真的值得换吗？我有点怀疑” Redditen机器翻译 展开原文I get around 100-200 t/s with qwen3.5-9b combined with dflash. Is it really worth switching? I'm kinda skeptical负面编程2026-09-03 “在编程方面，它确实比Qwen3.5 9b更好” Redditen机器翻译 展开原文It's legit better than Qwen3.5 9b in coding正面通用文本2026-08-29 “你甚至可以关闭vision来为3090节省空间，然后在9b模型上启用它——根据我的经验，这个模型能力很强，而且能让3090获得更多上下文。” Redditen机器翻译 展开原文You can even drop vision to save space on the 3090 and enable it on the 9b model which is very capable in my experience and allows for more context in the 3090正面图像/视觉2026-08-29 “同样的图表用qwen3.5-9b跑出来的结果是准确的，读取了所有数值。所以我为了保险起见又换回去了。” Redditen机器翻译 展开原文The same chart run through qwen3.5-9b was exact and read all values. So I swapped back just in case.负面编程2026-09-03 “根据他们的基准测试，7B在编程方面似乎优于qwen 3.5 9b，而在其他领域则持平或略逊一筹” Redditen机器翻译 展开原文7B seems to beat qwen 3.5 9b in coding while matching or falling slightly short in other areas according to there benchmarks正面推理2026-09-01 “在bs bench上得分很好，这意味着即使你在胡说八道它也不太可能一直同意你” Redditen机器翻译 展开原文scores well on bs bench, which means it's less likely to constantly agree with you even if you're talking nonsense 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
负面编程2026-09-05 “Qwen会提供备份和打印输出，但总是有1-3个错误” Redditen机器翻译 展开原文Qwen will offer backups and printouts but always has 1-3 errors
正面速度与延迟2026-09-05 “我可以和Qwen3.5-9b以及更小的模型进行有趣的对话” Redditen机器翻译 展开原文I can have fun chats with Qwen3.5-9b and anything smaller
正面速度与延迟2026-09-03 “我用qwen3.5-9b配合dflash大概能跑到100-200 t/s。真的值得换吗？我有点怀疑” Redditen机器翻译 展开原文I get around 100-200 t/s with qwen3.5-9b combined with dflash. Is it really worth switching? I'm kinda skeptical
正面通用文本2026-08-29 “你甚至可以关闭vision来为3090节省空间，然后在9b模型上启用它——根据我的经验，这个模型能力很强，而且能让3090获得更多上下文。” Redditen机器翻译 展开原文You can even drop vision to save space on the 3090 and enable it on the 9b model which is very capable in my experience and allows for more context in the 3090
正面图像/视觉2026-08-29 “同样的图表用qwen3.5-9b跑出来的结果是准确的，读取了所有数值。所以我为了保险起见又换回去了。” Redditen机器翻译 展开原文The same chart run through qwen3.5-9b was exact and read all values. So I swapped back just in case.
负面编程2026-09-03 “根据他们的基准测试，7B在编程方面似乎优于qwen 3.5 9b，而在其他领域则持平或略逊一筹” Redditen机器翻译 展开原文7B seems to beat qwen 3.5 9b in coding while matching or falling slightly short in other areas according to there benchmarks
正面推理2026-09-01 “在bs bench上得分很好，这意味着即使你在胡说八道它也不太可能一直同意你” Redditen机器翻译 展开原文scores well on bs bench, which means it's less likely to constantly agree with you even if you're talking nonsense
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。
1 条精选摘录
展开原文
Qwen will offer backups and printouts but always has 1-3 errors
展开原文
I can have fun chats with Qwen3.5-9b and anything smaller
展开原文
I get around 100-200 t/s with qwen3.5-9b combined with dflash. Is it really worth switching? I'm kinda skeptical
展开原文
It's legit better than Qwen3.5 9b in coding
展开原文
You can even drop vision to save space on the 3090 and enable it on the 9b model which is very capable in my experience and allows for more context in the 3090
展开原文
The same chart run through qwen3.5-9b was exact and read all values. So I swapped back just in case.
展开原文
7B seems to beat qwen 3.5 9b in coding while matching or falling slightly short in other areas according to there benchmarks
展开原文
scores well on bs bench, which means it's less likely to constantly agree with you even if you're talking nonsense
近 7 天暂时没有符合此筛选条件的精选评论。