I Munch LLM’s Boxes

@SIGKITTEN Damn guess I’ll wait until iOS update. Btw loving the updates happy I tried it out

@TokenMuncher on testflight yeah

you could be scrolling X and watching codex the movie in PiP

@ericzakariasson I wish I worked at cursor so I didn’t have to cuck to just OpenAI or Anthropic :(

i use this model exclusively for any ui work i might do

@TheRealAdamG @ivanburazin All these retards get paid for saying controversial shit it’s actually disgusting

@ivanburazin Have you tried GPT-5.5 or Codex?

I don't think OpenAI will come out on top in the coming years. Being first doesn't mean you last forever. By trying to be aggressive with growth and revenue, they have clearly overpromised, ventured into everything at once, and gotten distracted. They launched a bunch of products and experiments. None of them survived except ChatGPT. Meanwhile, ChatGPT itself hasn't changed in a year. Models feel lazier due to token optimization. Responses are also shorter and less thorough. Anthropic, on the other hand, stayed focused on coding. Coding became general purpose agents, and now they're clearly winning that market. But this also feels counterintuitive, since consumer-facing applications usually win out in the long term. - iPhone vs BlackBerry - GitHub vs GitLab - Apple vs IBM Common logic says whoever owns the end customer can move upmarket more easily than enterprises can move down. OpenAI still owns most of the consumer mindshare through GPT. In most countries, the average person hasn't even heard of Claude. But they're squandering away this massive advantage + the initial headstart.

@matt_feroz Talked to an agency it was 100k for 5M impression on x

Few know this strategy…

Yo why did @steipete block me wtf I miss seeing his shit :(

@alvinsng Are all the skills yall use for clean agent code oss?

Test-driven development, lint-driven development, and now: skill-driven development. We use Knip to catch dead code, but engineers started adding ignores when violations were "hard" to fix. So we built a skill that fixes these issues systematically. Now they just say: "use the Knip skill and fix it."

@SIGKITTEN 😬

shit im fucked

Rookie numbers

@jxnlco @elder_plinius Lmfao 🤣

agents unbanning agents 🤝

@SIGKITTEN @LLMJunky Wow this is wicked

This is awesome, on android, pets can live outside the app too Thanks @LLMJunky !

@yacineMTB You think these queers are gonna win now? I’m kinda torn. Was loving shitting on hocus pocus

lmfao

@nickbaumann_ Holy fuck that’s insane

for pretty much everything

@theo Lmfao

Here's our token utilization for those curious

@SIGKITTEN Or am I just retarded

@SIGKITTEN Nice this is legit. Onboarding could’ve been clearer to help me get signed into my gpt account and my mac :). Also no support for 5.5 models?

this is crazy

@0xSero How’s their plan hold up to max codex or cc plan? I love droid but my OpenAI proxy always has bugs with new model drops on their platform

I think it’s time for everyone to give Droid a shot. New usage system. - 100$ plan - usage pools - secondary usage pools with Droid Core (open weight models) I love this

@thsottiaux why is codex image gen hallucinating random unrelated images lol. 2nd time its happened

@FactoryAI Needed this last week when I took down prod checkout link 😭

The tests passed and the linters were green, yet the login form still broke in production. Why? Because nobody actually clicked through it. Today, we are launching Automated QA in Droids to fix this.