al3rez

I've made $70K building MVPs using Cursor (I am a giving a way my playbook) And I only paid: - $100 in Cursor tokens - $17 in @AnthropicAI tokens - $20/month for Claude/ChatGPT (Often one at a time) And: - @boltdotnew for full-stack development - @lovable for design to MVP development And they both cost us around $20/month + $50/month There is no better time than now to be a founder Comment "MVP" and drop a follow. I’ll DM it to you. P.S. This will likely blow up, so give me some time to reply.

@theo I just set up this on my Mac mini m4 as a sandbox before was using Paeso

It's still so, so cool that I can control all of my dev boxes through T3 Code. I barely open the desktop app anymore, I just use the site and the app. Working on 3 projects with 2 harnesses across 4 dev machines, all from one (open source) interface 🫡

@robj3d3 Welcome on board 🛹

Used Fable 5 to prep my entire Thai visa application 🇹🇭 > drafted cover letter > translated documents > structured financial evidence > reviewed everything against checklist 1 week later: approved ✅ 5 year visa babbyyyyy

@thdxr github.com/al3rez/zehn you can search through all of your agent history I always forget what agent I use for what so I made this

please i'm begging you show me something you built not another "this is my custom agent setup" post where you pretend you're doing something smarter than vanilla claude code please

@anthonyriera i really like the ux

@robj3d3 In Bangkok that'd be at least $600-1,000 baht depending on hospital

Also can you believe this? > 2x ECGs and a blood test > in and out in 1 hour > total $35 Thank youuuu Vietnaaam 🇻🇳 Europe would never 🇪🇺💩

@davidim roy lee is wining

we've hit 20k users in 8 weeks. ~95% of users came from organic content on instagram and tiktok for people in twitter : if u want to sell to people outside tech, learn to hack ig/tiktok

@GeorgeLampro20 lets gooo

Calm almost $3k day

@badlogicgames no no you have to write 100 .md files to improve the accuracy of the model lol

i get asked what memory system i use often. my answer has always been: my codebase is my memory system.

@ky__zo looks sick

while everyone was freaking out about fable and sol, I took my first week away from the laptop since 3 years i'm back today and guess what - fable is still here, world didn't collapse and i have a new core memory with cows 🐄⛰️ gotta keep touching grass more often. Italy's so pretty 🇮🇹

@robj3d3 @AnthropicAI take care bro

I'm done with them fucking with us. Ended up in hospital today from stress. Stayed up all night pushing my limits too hard, thinking it would be removed. Health comes first. Do better @AnthropicAI

@antirez how you package your idea into a sellable product is more important than execution now.

Worth stressing.

@WidmerLuiz @theo yup luna max all the way

@theo Gonna be the boring one here, technically 5.6 Luna at MAX reasoning already surpasses 5.5 xhigh in many benchmarks. The clear path for replacement is Luna. The others are glorified token burners. I do spun up Sol sometimes I like it, but the daily driver is Luna.

@theo Currently Luna max is the only thing is valuable to use for almost anything for a lot less token/hitting rate limit

Fable 5 on low already matches the best Opus 4.8 can do, at roughly half the cost per task

@filicroval it's cool but unreasonably complicated

guys hear me out: new pachinko reasoning picker for Codex 💸

AI news were so exciting in 2022-23, then turned into a jungle where 95% is "noise".

@cjayls @skirano Yes but less credit so basically Luna Max can do all with more time but less token

@skirano But when comparing luna max / terra max to Sol high / xhigh, don’t they take much longer?

Built a site to explore the Coding Agent Index better, with a few surprises: Terra Max edges Fable 5 Max (77.4 vs 77.2) for ~76% less per task. Sol XHigh is 1 point behind Max at ~26% lower API cost. Luna Max beats Opus 4.8 Max for ~80% less per task. 👇

@skirano I am using Luna max for everything and Fable 5 specific things

@Enrique_oriol a demo and how does it work?

#Fable5 was out, so I had to test it. My weekend AI workout: Set up Claude Code in my server + LLM Wiki. I've been wanting an AI-augmented dev environment running 24/7 for ages, and it finally happened! Here's how I integrated Claude in my VPS with mobile access 👇 #BuildInPublic

@robdel12 @antirez what font is that?

@antirez I still read the code because models do dumb shit like this regularly. I couldn’t imagine just letting the slop run away

It is my belief that many devs right now are not maximizing what they can do with automatic programming because they still look at the code. Doing it makes you the bottleneck. Your time is better invested in new ideas, QA, design, and asking yourself what is your goal.