LDG

[LDG @donggeng_li]

@Davidwuuu92 @LDLoeb @thsottiaux 200 刀的跑 Astra，复杂的一个任务都跑不完，几个小时就耗完了

[LDG @donggeng_li]

@Lucius_prps @0xjackkk 咱俩同样的配置，我的昨天用 Fabel5.1 的时候被封了，稳定用一年的号

[LDG @donggeng_li]

@xluchuan 我昨天刚被封，申诉了不予回复

[LDG @donggeng_li]

@LZRationalnvest 我没有提示让用手机号注册，用国内 Visa 卡绑定 Google Pay 付款的，已经稳定用了 3 个月了，也很奇怪为啥不封，我都不敢开最高的 Max 套餐, 现在还在用x5

[LDG @donggeng_li]

@AI4Azure 做中转站吧

[LDG @donggeng_li]

@_mohansolo I’m never using it again.

[LDG @donggeng_li]

@dashboardlim I need it

[LDG @donggeng_li]

@CTracy0803 在MacroAI 的榜单里面，Gemini 3.5 Flash 甚至能排到第一名

[LDG @donggeng_li]

@OfficialLoganK @andonlabs Pareto optimality doesn’t mean the model is the most usable, let alone perfect. It only means you can’t improve one area without hurting another metric. As an Ultra user, that’s my biggest concern, and I’ve already canceled my subscription and switched to GPT Pro.

[LDG @donggeng_li]

@Amank1412 I am an Ultra user. I feel that Antigravity is clearly inferior to Claude. Although it features a built-in Claude model, the usage quota runs out quickly. Their own Gemini 3.1 Pro is even more useless and poorly polished. Don't buy it. I am planning to switch to Codex later.

[LDG @donggeng_li]

@theonlysol_ I am an Ultra user. I feel that Antigravity is clearly inferior to Claude. Although it features a built-in Claude model, the usage quota runs out quickly. Their own Gemini 3.1 Pro is even more useless and poorly polished. Don't buy it. I am planning to switch to Codex later.

[LDG @donggeng_li]

@antigravity Could you please optimize your product first? There used to be all sorts of error codes, and now it doesn't even show those anymore—it displays absolutely nothing after running.I've tried restarting and opening a new window, but neither worked.

[LDG @donggeng_li]

@bitfool1 确实，我也是Gemini生成又快质量又高，而且情商也很高

[LDG @donggeng_li]

@LipuAIX 比atlas强在哪里？

[LDG @donggeng_li]

@AIPMLogi 这价格是我宽带费的几倍😂，我是买了几个便宜的，设置好策略交替用

[LDG @donggeng_li]

@jesselaunz @bao_xiao78791 文稿匹配一次只能5000字，有没有什么好的解决方法？

[LDG @donggeng_li]

@CitrusFurina @seclink 我的也是，12核的m3 pro卡的一逼，还是得用GPU

[LDG @donggeng_li]

@dotey 我怎么感觉claude3 sonnet没gpt4 turbo好用呢，同样的代码生成任务还是gpt更准确，claude优势是确实很快，而且超长的上下文

[LDG @donggeng_li]

@geekbb 不如coze

[LDG @donggeng_li]

@knowledgefxg github上的v2ray就很好用