Lilo

nonono.

[Lilo @Lilo]

@thsottiaux Explain about "gpt-6-astra-degrade2-luna-1p-codexswic-ev3" ? I asked for the flagship. I paid for the flagship. Silently swapping in a cheaper model at full price isn't "optimization" — it's fraud.

[Lilo @Lilo]

@thsottiaux Explain about gpt-6-astra-degrade2-luna-1p-codexswic-ev3? I asked for the flagship. I paid for the flagship. Silently swapping in a cheaper model at full price isn't "optimization" — it's fraud.

[Lilo @Lilo]

So what does OpenAI have to say about that? Actually, never mind—I don’t care anymore. If this is how paying users are treated, there’s no reason for me to stay. No transparency, no accountability, no standards. (3/3)

[Lilo @Lilo]

Canceling my OpenAI subscription and farewell Codex. The silent degradation in service quality—model downgrades, throttling, and worse output—has become completely intolerable. I’m done paying for a service that behaves like this. (1/3)

[Lilo @Lilo]

Contacting support goes nowhere: I just get boilerplate replies insisting everything is fine. Meanwhile, the community is now full of technical write-ups showing ways to partially bypass or mitigate these apparent quality degradations.(2/3)

[Lilo @Lilo]

@Francisco7flf1 不是，而是这傻逼给啥订阅都降，pro 20x 直连不反代都被降。

[Lilo @Lilo]

@thsottiaux Just stop secretly degrading model performance or quietly routing users to vastly inferior models. For people paying real money for a subscription, deliberately lowering service quality and concealing that degradation is fraud.

[Lilo @Lilo]

@thsottiaux Preparing to request a refund and move to Claude Code instead, good work.

[Lilo @Lilo]

@thsottiaux Stop secretly degrading model performance or quietly routing users to vastly inferior models. For people paying real money for a subscription, deliberately lowering service quality and concealing that degradation is fraud.

[Lilo @Lilo]

@linglingfa 非常简单的改改文档，探查代码（还不能做详细原因分析，得是简单那种）。感觉最大的用途是反代出来给沉浸式翻译🤣

[Lilo @Lilo]

I used to dread “can you turn this into a PPT by EOD.”

Now I just throw the docs into ChatGPT Work and it builds the deck + the write-up from my actual files. Still need to polish, sure—but the blank page is gone.

[Lilo @Lilo]

@Jonathan_Blow macOS, please! I really don’t want to have to stream it anymore.

[Lilo @Lilo]

@Jonathan_Blow It’s been about two weeks! What’s the new Order of the Sinking Star announcement? I can’t wait to hear more 👀

[Lilo @Lilo]

@Jonathan_Blow I'd suggest taking a careful look at the contractor's localization work. Please don't let this end up like the Braid Anniversary Edition commentary track, where the Chinese translation was such a mess that it seriously detracted from the experience. Thank you.

[Lilo @Lilo]

@Jonathan_Blow Seeing this screenshot, I'm starting to worry about the Chinese localization.

Based on these two lines alone, it feels less like a proper translation and more like raw machine translation. The wording is awkward enough that a modern AI translator would probably hive done better

[Lilo @Lilo]

@FeloApp Followed & RTed!

[Felo Translator AI Real-time translation |Official @FeloApp]

＼2024 Work Efficiency Support／

Introducing the 【Felo Translator】, AI Real-time Translation App!
Act fast, as we've prepared 1️⃣0️⃣0️⃣ referral codes on a first-come, first-served basis!

To get a code: Follow and RT!

Each code comes with 6️⃣0️⃣ minutes of usage time!

[Lilo @Lilo]

@Tisoga @threadreaderapp
unroll

[Lilo @Lilo]

@Neil_Druckmann As long as this m*f*er is still in, I will never spend a penny to Naughty Dog again

[Lilo @Lilo]

@kusatta_www @shipinbot www