Damn! I missed this offer. Can someone from @claudeai share this again? I will subscribe to Claude Max 20x plan if you do 😁😁😁

We moved all our apps and services to @coolifyio last night. Apps: 1. Next.js App (Preview and Production) + Auto Preview Deployments 2. Sendy 3. Plunk (Testing) 4. Fuma Docs (TanStack Start) 2 PostgreSQL 18 Databases RustFs (S3 Storage) Overall, we are happy with the migration. Since the Database and App are on the same server, it reacts much faster. We reduced our costs by 80% with almost zero maintenance overhead. Thanks @coolifyio : )

@GeminiApp Team, it hurts! Why can't you build a good model

While the world is using Opus 5.5, here is how @GeminiApp responds to every task, big or small. It loves to overcomplicate things, even when you instruct it to not do it. I am tired.

3. Hallucinates and makes up information a lot of the time. You can see similar behaviour in Google AI search chat, where, if you ask it something, it just makes up information instead of saying I don't know. This hurts especially when doing code reviews, as the findings are bad and I just default to GPT Sol for review.

I have been using the new Gemini Flash 3.8 model for a month now. The model is decent, so I have paired it with GPT Sol. I would ask the GPT model to generate a proper plan and then hand off the task to Gemini. Surprisingly, it works decently well, but I did face a lot of issues. Overall, I would say if I am using a Gemini model, then I am not confident about its output and will always double-check and find something weird. A few issues I always face👇👇

Why don't you reset our Cursor/Grok limits?

I have been trying @trySynara, and it mostly works fine. But I think @emanueledpt has to focus more on stability than adding features at this point. Or have a separate preview version for people who want to use an unstable version. Here are some of the issues I encountered: 1. My Claude chats kept disappearing as I added new messages. 2. After a point, Codex keeps stopping. I have to reply with Continue, but it just doesn't work. (Maybe a Codex issue.) 3. The thread remains in a running state, and the Stop button doesn't work. 4. Cursor doesn't work. Earlier, I used to get a Model ID mismatch error. 5. Revert changes seem broken. Maybe 2 in 10 times it worked for me.

I used ~70% of my Cursor $20 usage in one day. @cursor_ai I am not sure whether the Cursor $60 plan can last longer. And the big question is how long? Model: Grok 4.5 I can do much more work using the $20 Codex/Claude plans. Btw, I have three $20 plans.