
🇺🇸 Software Engineering at Jala University 💬 CCO at @berolabx 🔍Building https://teapawt.org/t.co/jDD9EIr0lj 🃏 Indie hacker
São Paulo - SP
Joined June 2025
- Tweets1K
- Following123
- Followers303
- Likes1.6K
Pinned Tweet
Usei o claude enterprise do meu trabalho pra fazer meu SaaS usando o opus no talo, ainda bem que não da pra ver essas coisas no painel de admin.
Se quiser conferir, link do beta amanhã aqui nos comentários
E se o seu time de QA pudesse automatizar testes sem perder o controle sobre eles?
Esse é o propósito do 𝗤𝗔𝘇𝗶𝗻𝗵𝗼.
Um assistente de QA que explora aplicações, executa testes de ponta a ponta e encontra falhas em web, Android e iOS.
É 𝗤𝗔𝘇𝗶𝗻𝗵𝗼. É 𝗦𝗮𝗹𝘃𝗲 𝗦𝗼𝗳𝘁𝘄𝗮𝗿𝗲.
Link nos comentários 👇
cc: @sseraphini @meunomeebero @ducaswtf @BolhaDevCom
dann retweeted
🔥 中午刚说 40 系还没支持,今晚 4090 到 4060Ti 全解锁了,连 FP8 都翻倍
40 系不能直接套 30 系的模块,是因为它换了另一套签名引导固件,多锁了四道权限门。今天研究者 TOTORONG 放出 AD10x 载荷,在驱动绑定前一次捅开四道门,而整个 40 系共用同一份引导固件,所以从 4090 到 4060 Ti 一下全通了——同样不刷 vBIOS、不改驱动。
1️⃣ RTX 4090:FP16/BF16 指令吞吐从 165.6 涨到 331.6 TFLOPS;500W 下跑 cuBLAS 8192³ 矩阵乘,TF32 从 77.5 涨到 165.9 TFLOPS(2.14 倍),3 张非公卡已复现
2️⃣ RTX 4080 SUPER:PyTorch 矩阵乘 FP16 从 110.6 涨到 202.7 TFLOPS(1.83 倍);mmapeak 实测连 40 系独有的 FP8(FP32 累加)都从 201.8 涨到 376.1 TFLOPS
3️⃣ RTX 4070 Ti SUPER:240W 下 FP16/BF16 从 95 涨到 189.9 TFLOPS;跑 CUTLASS 4.8.1 原生 FP8 快累加矩阵乘从 158.5 涨到 309.1 TFLOPS(1.95 倍),322 万个输出单元零比特差异
4️⃣ RTX 4060 Ti 16GB 与 Win11:三卡同机 A/B 实测,182W 功耗墙下 BF16/FP16 仍涨 38–39%;Windows 11 也出了免核显、单卡直接亮机的 UEFI 引导版
💡 Linux 端装下方仓库(4080S 看原仓 PR #24)。40 系别用 targets=all,先跑 python3 tools/print_boot_option.py 拿到单卡参数加进内核启动项,关掉 Secure Boot,关机拔电源等几秒「冷启动」才生效;Windows 端去 Hugging Face 搜 TOTORONG/AD103_win_unlocker。
github.com/JustAResearcher/g…
中午还在等 40 系的各位,今晚敢拔电源试一把吗?
dann retweeted
Confirmed that 3090 unlock is successful
with 360watt power wall, I got 75T -> 135T (measured in Pytorch)
Which is insane
I used github.com/Zagitta/gpu-unloc… with commit a9b0c75
dann retweeted
Vão assistir o vídeo do mano dan falando sobre jev no canal da berolab agora mesmo ou vire uma lesma condenada a passar a eternidade se alimentando de cuscuz paulista youtube.com/watch?v=wLu8vcf-…
they’re calling it “Black October” and i don’t think people realize how serious this is
in the last 48 hours there have been over 27 major cybersecurity incidents. not in a month. not in a week. in 48 hours
any ONE of these would normally be the biggest story of the year
PHP: critical vulnerability in the software running 80% of the web. severity 9.2 out of 10
Node.js: high-severity flaw letting attackers take over servers remotely. affects versions 22, 24, and 26
Java: also compromised. three of the most used programming languages on earth hit at the same time
GitLab: a perfect 10 out of 10 severity hole in every version since 2014
Railway: hosting platform compromised. API keys for OpenAI, Claude, Stripe potentially exposed
Signal: a bug that could leave your mic and camera on after you hang up
WordPress: hackers stole a plugin maker’s keys and planted backdoors on websites
LiteLLM: an AI tool downloaded 3.4 million times a day was poisoned with a password stealer
and that’s not even half the list
thousands of AI agent swarms are reportedly scanning and attacking vulnerable servers right now. automated. around the clock
security researchers are saying this might be the early stages of a global cyber incident that nobody fully understands yet
if you run a website, use cloud services, store API keys on hosting platforms, or use any software built on PHP, Node.js, or Java you are potentially affected
update everything. rotate your API keys. change your passwords on critical services. enable 2FA everywhere you haven’t already
this is not a normal week on the internet. this is something different
An urgent notice from the Railway team
blog.railway.com/p/security-…
dann retweeted
i just shipped a landing page for MacStats
check it out: macstats.luizcastro.dev
if you’re on Mac, download it, give it a try, and let me know your feedback
Primeiro vídeo da série postado! Espero que gostem. 100% feito com código, agora é melhorar.
youtu.be/akt0UFrNlLs?is=tdd4…
#bolhadev
72
dann retweeted
Foi um ótimo recomeço para a @berolabx . Mais uma vez, obrigado @lincolixavier por estar sempre compartilhando teu conhecimento conosco.
Essa foi uma call aberta, quem perdeu ainda pode ouvir a gravação. acessa: berolab.com/
cc @meunomeebero @berolabx @danndevIH
Rolando a callzinha braba na @berolabx com o @lincolixavier
O HOMI DO MELHOR LANCE!!!
Bora que dá tempo de entrar:
berolab.app/meetings/01a07c4…
dann retweeted
I resigned from Anthropic today. I spent the last three years doing pretraining research at both OpenAI and Anthropic. Neither company is acting responsibly. They are racing straight to self-improving superintelligence and gambling with our lives. More thoughts below.
dann retweeted
Em meio a tanto conteúdo de IA, eu tentei aprender haskell em 30 minutos com meu amigo que não sabe haskell
é trabalho simples mas é honesto
GLM-5.3 × ZCode Weekend Build 🚀
Eligible new users: log in to ZCode for the first time from Aug 16, 00:00 to Aug 17, 09:00 (UTC+8) to automatically get 100M free GLM-5.3 tokens. Limited availability; ZCode only.
Try it: zcode.z.ai
🤖 Made with AI
dann retweeted
Only if education could be this interactive ❤️🔥
I've had a looong wish to build something genuinely useful through vibe coding, and I finally did it.
A 3D human anatomy application built with @threejs using GPT 5.6 Sol.
It all started with a single design image that I created using GPT Image 2.0. I then used it to generate every 3D organ image, one by one.
Next, I converted each of those images into 3D models using @tripoai (and no, they didn't sponsor this 😄).
After that, I opened Codex, wrote a master prompt based on the design, and gave it the prompt, the design image, and all the 3D models.
Codex built the first version beautifully, but there was one big problem.
Every single 3D model was nearly 120-150 MB. That obviously wasn't practical for the web and was giving a performance of 16fps.
After a few iterations, Codex optimized each model down to roughly 2–5.5 MB while preserving the visual quality, reducing the total asset size from ~900 MB to just 28.6 MB. And each model loads on demand.
Along the way, Codex also generated those anatomical illustrations showing where each organ sits in the human body, and even created the interactive hotspot markers that explain different parts of every organ. It handled all of that.
The process wasn't exactly one shot, but it also wasn't difficult. You just have to do it step by step.
It genuinely felt like building something that could make learning anatomy much more engaging.
The inspiration came from @DilumSanjaya's 3D animal plant cell project. I remember seeing it and thinking, "I want to build something like this one day."
And I did it :D
Live: anatomy-livid.vercel.app/
Code: github.com/thebuggeddev/anat…
dann retweeted
Back at the laptop. The usage limits have been reset for all paid users of Codex and ChatGPT Work. Weeeeeeeee. It's a good day!
dann retweeted
We’re celebrating the fast adoption of chatGPT Work and all the incredible effort that went into it today. I’m feeling like a limit reset.
Hold on tight to your ultra and /fast and see you in a few hours when I’m back at the laptop!



























