Thilak Rao

@rudrank I did, and the author was a complete asshole who refused a new license or a refund, even though I’d only had it for two weeks. Good luck shilling griftware, but if you ever get sick of paying $100 a year, it’s just one prompt away.

@Thilak I assume you would have reached out to support for it; anyway enjoy your vibe-coded stuff

This lad is shilling Astro for affiliate commissions, but the reality is the product is a complete scam. You have to bring your own API key, and the license is tied to a single computer and user. I bought a yearly subscription to this griftware, then upgraded my Mac and lost my license. It took me one afternoon to vibe-code this. My fork now also pulls financial data from ASC and search data from GSC.

1. Benchmark Claude Opus 4.6 against an OpenAI model that's three generations older (GPT-5.2-Codex, GPT-5.3-Codex and the latest GPT-5.4 are all newer than GPT-5.2). 2. Declare that Claude Opus 4.6 is the best model for vibe-coders. 3. AI thot leaders poast, repoast and their gullible vibe-coding audience lap it up, hook line, and sinker. Open weight models bad, OpenAI models bad, Claude is the bestest model ever! /s I know it isn't a panacea, but this field desperately needs strong disclosure of interest policies, the way more established fields like medicine have. nb: They have models newer Chinese models from 02/2026 in their benchmarks, so this almost certainly isn't an error of omission.

@zkyo @JasonBotterill Codex $20 plan is worth more than Claude Max $200 plan in usage.

@JasonBotterill Claude offers both $20 and $100 options. In contrast, OpenAI jumps directly from $20 to $200.Personally, I don't have such a high usage volume, so for now, I'm sticking with the $100 Claude Max plan.

The ChatGPT Pro plan gives way more value than Claude Max now. GPT-5.4 thinking is high taste and unlimited, much more Codex limits, and you actually get /fast mode as part of the subscription. Anthropic doesn’t let you use /fast mode even on the $200 tier.

@solipsnitsyn @JasonBotterill Not really. With Claude, I hit my Pro limits within a couple of prompts. That’s not the case with Codex. You really have to try hard to max out your usage.

@JasonBotterill tbf nothing on pro is unlimited. you'll hit abuse guardrails and then end up sitting there for 2 days being silently routed with absolutely no notification or heads. it's why I moved to ai ultra, beyond just offering a better suite of tools in general

@JasonBotterill The only reason Claude Max plan is still $200 is @OpenAI 🤣

ADHD multitaskers absolutely printing rn > talk to agent 1 > while you wait for it to respond, talk to agent 2 > while you wait for agent 2 to respond, back to agent 1. if still spinning, talk to agent 3...

@srdjanbogi Codex CLI does :)

@Thilak Does it have any kind of planning mode, like CC?

The $20 ChatGPT plan with Codex has been a much better deal than Claude Code for my day-to-day usage. I usually hit Claude Code’s weekly limits in a day or two. This weekend I put Codex through hours of work and I’m not even close to the cap. GPT Codex 5.3 (Extra High) also feels more reliable than Opus 4.6 for coding tasks: fewer errors and less back-and-forth prompting. The main downside is speed, but I’ll take slower responses over more retries.

Codex app finally made Git worktrees feel smooth. My flow is simple: - I write prompts in the Codex app. - When I need more control over code or the terminal, I hit Open and jump into Antigravity. Worktrees are the key. Each long-running task gets its own isolated checkout, so I can run multiple threads in parallel. It’s especially helpful if you work in a monorepo. And with @OpenAI temporarily offering 2x Codex app rate limits, this setup has been letting me run more parallel tasks and get more done.

@andrewxroas @mr_vibe_it Because it’s fake? The account is verified since less than a quarter with one username change in Jan 2026. Best is to show proof of payment, but no… Highly sus

@mr_vibe_it I've shown proof countless times Sorry to break it to you but no one who's doing primarily paid UA will share their app

Finally reached 500k MRR with my app Most companies never end up reaching this scale We did it in less than a year Time to take this blueprint and scale multiple apps 2026 will be insane

@Teknium I use Claude Code with GLM 4.7 because Claude Code is too expensive. I can’t tell the difference between GLM 4.7 and Sonnet 4.5

Openai is in trouble now

@shoryaDs7 @WSJ Try Claude Code with GLM 4.7. The best $3 alternative to Claude Code Max plan.

@WSJ Claude Code hype is real until tomorrow: forgets the whole project, asks who you are. Then one loop eats your tokens and boom, limit hit. Feels like premium price for a forgetful, expensive goldfish 😭

They call it getting “Claude-pilled.” It’s the moment software engineers, executives and investors turn their work over to Anthropic’s Claude AI—and then witness a thinking machine of shocking capability, even in an age awash in powerful AI tools. on.wsj.com/4k2CU9E

In 2024, normies didn't understand that there was no way to reliably detect AI-generated writing. In 2026, AI users don't understand that there *is* a way to reliably detect AI-generated writing. And also that readers have developed a nose for the stink of AI prose!

@SgtRock4445 @jmscaronte @grok .@SlopOrNotAI says it’s fake

@jmscaronte @grok is this picture real or ai?

Bueno,noche intensa en Venezuela,intentare subir lo que encuentre de la misma.Empezamos con rueda de confirmando operativo sobre el pais. #Venezuela

@jmscaronte Looks fake

La foto del año…

@conciente15 @LillyTellez @grok Nope, it’s fake

La justicia llegó tarde, pero llegó. Que esta foto sirva de ejemplo para los narcopolíticos mexicanos, cómplices del tirano Nicolás Maduro. Los venezolanos tienen derecho a ser libres.

@LillyTellez This is a deepfake. I used @SlopOrNotAI to check

@JassxRandhawa @Finallydiditonx @smitaprakash @SlopOrNotAI

This picture of Maduro (not sure if it has been AI enhanced) posted by OSINT sites seems a tad too similar to the capture of Saddam Hussain.