
hi, i'm Deepu cs student who loves using AI and talking about it. building way too much stuff, hopefully shipping something cool soon also making fun stuff every now and then. stick around

I like talking about ai and building things. all I really want is the freedom to keep doing what I love.

hi, i'm Deepu cs student who loves using AI and talking about it. building way too much stuff, hopefully shipping something cool soon also making fun stuff every now and then. stick around

I bought a plan because I was really excited to try @capydotai, I've heard great things about them but I'm pretty disappointed 2 of the 3 models they call unlimited free are actually "Free with Ads" and "US and Canada only", and neither is mentioned on their pricing page or in their docs I also missed the trial. my first card didn't go through, so I switched to another one and it charged me for the full plan instead. not a big deal but just wanted to point out the bug it's not about the money, but not mentioning this before you pay is a really bad first impression. I'm sure a lot of people will buy the plan for those models, so it should be clearly mentioned in the docs also the two rows at the top are VERY annoying. they take up too much vertical space and I don't think there's a way to turn them off. I'd like to at least hide one this pissed me off way more than it should have idk why, but I know you guys are great and will probably fix most of this soon 🙏 cc @justinsunyt @xPolarrr
We just made agent swarms affordable. Introducing new Capy plans with up to 300,000 cloud VM minutes and 2,500 reviews, starting at $1. Plus 3 unlimited free models: Particle-1, DeepSeek V4.1 Flash, and GLM 5.3 Flash. No sub required. Comment below for one month of Capy Lite.

@Screeendev do uk anything about ts


hmm I'm in, looks nice, is it safe to use my claude sub in Capy? btw GLM 5.3 Flash and Deepseek V4.1 flash are "free with ads" and *Us and Canada only* :)
wtf are you high on @justinsunyt this is just crazy 💀 and exciting I was feeling weird and bored, Imma try Capy now

I did not know you could generate a new debit/credit card online and get it instantly gawd damnnnnn, bless my bot 🙏


wtf are you high on @justinsunyt this is just crazy 💀 and exciting I was feeling weird and bored, Imma try Capy now
We just made agent swarms affordable. Introducing new Capy plans with up to 300,000 cloud VM minutes and 2,500 reviews, starting at $1. Plus 3 unlimited free models: Particle-1, DeepSeek V4.1 Flash, and GLM 5.3 Flash. No sub required. Comment below for one month of Capy Lite.

thinking of getting this mic, wdyt anyone used it? or any better recommendations?


T3boat is here, gonna test it and share my experience here great work by @paukraft
My laptop can only handle so many @t3dotcodes agents at once. So I built t3boat: one click creates a cloud machine on @boatdotdev and adds it to T3 Code. Idle machines sleep, so you only pay for the minutes they run. Open source → t3boat.paukraft.com

don't fall for it, only download ardia through ardia.dev
🚨 Public service announcement There is a site trying to impersonate & trying to charge for ardia, and most likely malware infused download links fake: ardiabrowser (dot) lol original: ardia (dot) dev pls be aware and DO NOT CLICK DOWNLOAD OR PAY and if someone can help take this site down pls dm me. ardia is free to use and opensource Thankyou for your attention to this matter.

not feeling very poasty today, idk why lol there's so many things to do and I don't even remember half of them @bot helpp


you will see two types of posts today: > "there is a campaign going on against OpenAI and all these people saying bad things about Astra are paid" > "new Gemini whtvr periodic table model benchmarks leaked and mogs everyone, omg Google is so back 🙀" both are dumb, ignore and move on

I've been using different model family reviews for a while now, although without scoring. recently added scoring after watching Theo's video :D something I've noticed too is models getting stuck in loops, just patching things and never reaching an endpoint. oai models are the worst at this if you leave them unchecked :) something that worked for me (not for oai models tho, they love fixing everything and turning a 1k lines PR into a 100k lines) is adding an instruction like this: "if the council cannot converge, or keeps making patches every round and gets stuck in a loop, stop after X rounds (user set limit), zoom out and reconsider the approach. start a new council round to get their opinions on what approaches they would take, pick an approach the council can converge on, and continue from there." wdyt @theo