用户Nick_D和Ruthvik在Google AI开发者论坛发帖，请求不要停用Gemini 2.5 Flash模型。他们表示，该模型在其工作流程和内部基准测试中表现优异，而新版本Gemini 3 Flash和3.1 Flash Lite在性能上均无法替代，希望保留该模型。

近日，在谷歌AI开发者论坛上，多位用户发帖呼吁不要停用Gemini 2.5 Flash模型。用户NickD首先感谢了Gemini团队提供的优质模型，但指出其内部工作流程高度依赖2.5 Flash，因为内部基准测试显示，即使调整了提示词，新推出的Gemini 3 Flash表现仍不如2.5 Flash，且难以简单替换。用户Ruthvik也证实，在延迟和性能上最接近的3.1 Flash Lite远不及2.5 Flash，后者是全能型最佳模型，并推测该模型占据了大量流量和用户，希望谷歌能长期保留。用户JoshuaSimpson强调，2.5 Flash是唯一在澳大利亚部署的低延迟模型，能实现300-400毫秒的响应，适合语音代理；而3.5 Flash延迟高达600-700毫秒，且未在澳大利亚部署，实际延迟接近700-800毫秒，完全破坏了语音代理的使用场景。用户tylertreat则担忧成本问题，指出从2.5 Flash升级到3.5 Flash，成本增加了约3倍，认为Flash系列本应保持低延迟和亲民价格，但新模型定价偏离了这一初衷。用户Bcoun表示，如果2.5 Flash停用，他们将转向开源模型。用户merc也附和，称其关键工作流中，没有其他模型能在同等价格和智能水平上达到2.5 Flash的效果，停用将造成巨大损失，他们可能也会转向开源模型，但2.5 Flash的低延迟难以被超越。

以下是对评论内容的总结，涵盖主要观点、论据及不同立场，并保留了关键引用（中英文）。

1. 对Gemini 2.5 Flash被停用的普遍不满与惋惜

多数评论者认为该模型在性价比、速度和任务适配性上达到了“甜蜜点”，停用令人失望。 - 关键引用： - "It's such a good model for the price... outperforms gpt5 at 3x the speed and 1/5 the price." (mips_avatar) - "This is one of the best models you can use in an API... the perfect trifecta of fast, cheap, and smart enough." (kilroy123) - "Gemini 2.5 Flash is well balanced model that meets sweet spot of price vs performance trade-off." (ioreader)

2. 对Google频繁停用产品的批评与不信任

评论者指出Google有“杀产品”的历史，并警告不要过度依赖其云模型。 - 关键引用： - "Why does Google constantly kill off good things?" (kilroy123) - "Isn't asking Google to not discontinue a product a bit like asking the tide to not rise?" (Hamuko) - "How about you stop relying on Google products? You've learned nothing after all these years?" (throw_m239339)

3. 对成本上涨的担忧

从2.5 Flash到3.5 Flash，价格大幅上涨（约3倍），违背了Flash系列“廉价、低延迟”的初衷。 - 关键引用： - "I am more concerned about the cost step up... the latter being roughly 3x more expensive." (tylertreat) - "the era of cheap and plentiful AI might be coming to an end..." (tylertreat) - "gemini 3 flash is expensive... really not liking google these days they are not hungry anymore." (zuzululu)

4. 建议转向本地模型或开源模型

部分评论者认为，避免被“抽地毯”的唯一方法是自行托管模型，或使用开源替代品（如Qwen、DeepSeek）。 - 关键引用： - "If you use a local model discontinuation is no longer a thing to worry about." (segmondy) - "If they don't want to host it maybe they could open source it." (leumon) - "There will be such a massive shift to Qwen VL when Google shoots itself in the foot." (yieldcrv)

5. 对模型依赖性的反思与替代方案

一些评论者质疑过度依赖特定模型的做法，并指出已有替代品（如Gemma 4、DeepSeek V4 Flash）在性能或成本上更优。 - 关键引用： - "I have switched to gemma-4-26b-a4b... scores higher and is faster." (swe_dima) - "I've settled on deepseek-v4-flash as a replacement. Results are just as good." (ryukoposting) - "This is why I believe, for a company, to never be reliant on closed-weight models." (Alifatisk)

6. 对模型“个性”与主观体验的担忧

部分用户担心新模型无法复现旧模型的“写作风格”或特定任务表现，尤其是非客观可测的方面。 - 关键引用： - "how does one judge whether a new model is adequately reproducing the subjective 'writing style' of an old model?" (whycombinetor) - "I haven't found another model that performs as well for the specific tasks I was doing with it." (ryukoposting)

7. 对模型存档与长期可访问性的呼吁

有评论者提出，公司应在模型停用时公开权重，或由Archive.org等机构保存旧模型。 - 关键引用： - "If a company deploys a paid AI model and makes people depend on it, they need to dump the weights at EOL." (avaer) - "I wonder if some day we might see Archive.org organizations preserving older models as operating costs go down." (uproarchat)

总结：评论者普遍对Gemini 2.5 Flash的停用感到失望，认为其性价比无可替代，并批评Google的“杀产品”文化。同时，他们警告不要过度依赖云模型，建议转向本地或开源方案，并呼吁模型停用时公开权重。部分用户已找到替代品（如DeepSeek、Gemma），但仍有对模型“个性”和成本上涨的担忧。