I'm using MiMo Open Platform — Xiaomi's most powerful AI (MiMo V2.5 etc). Sign up with my invite code: both get $2 API credits 10% off your first plan. Invite code: 4VKGTR. Sign up: platform.xiaomimimo.com?ref=… (Auto-filled on sign-up · Credits valid for 40 days)

GLM 5.2 coding lite plan gives in 5 hour window

Like this postabl.xyz

Guys, tell me the best model which India have ever created.

openrouter.ai/docs/guides/fe…. . Any harness that can do this thing?

"What happens to Corporate Employees After 45? | Indian IT jobs after 45" By Varsha Kadyan (The Corporate Diaries) --- THE CORE ARGUMENT: The "work till 60, retire happily" plan is a myth for Indian IT workers. It's fundamentally flawed because it was borrowed from the previous generation where jobs were guaranteed for life. That no longer applies. --- KEY STORY: The speaker was in a meeting with management discussing appraisals. The CFO casually said about a senior employee (8-9 years at the company): "His salary has reached what we'd pay someone locally in North America. It doesn't justify keeping him." Nobody on the call pushed back except the speaker. The CFO's response: "That's for HR to solve. The numbers don't justify." This moment revealed the brutal truth about how companies view Indian employees. --- WHY THE PLAN IS BROKEN: 1.⁠ ⁠You were hired for being cheap, not exceptional. The whole reason Indian IT workers get work from global companies is because they're cheaper than local hires. That's the deal. 2.⁠ ⁠Every raise closes the gap. As your salary grows, the cost difference between you and a local hire shrinks. The moment that gap closes, you become replaceable — no matter how talented you are. 3.⁠ ⁠Your value has an expiry date. Companies never announce it, but the day your salary matches a local hire's salary, you're done. 4.⁠ ⁠Previous generation had guaranteed jobs. Our parents' jobs were secure until退休. Today, companies actively look for cheaper alternatives. 5.⁠ ⁠By 60, we'll still have debts. Unlike our parents who were debt-free by retirement, we'll have home loans, kids' education loans, etc. --- WHERE DID ALL THE 50 YEAR OLDS GO? •⁠ ⁠Forced early retirement (not voluntary) •⁠ ⁠Consulting roles (project-based, uncertain income) •⁠ ⁠Small businesses nobody believes in (no other option) •⁠ ⁠Moved to hometowns to slow down They didn't thrive — they disappeared. --- WHAT SHE DID: •⁠ ⁠Stopped fighting the system (negotiating hikes, competing for promotions) •⁠ ⁠Accepted the ceiling is structural, not personal •⁠ ⁠Started slowly exploring options outside corporate •⁠ ⁠Didn't quit immediately — explored gradually --- HER ADVICE: •⁠ ⁠Don't quit your job immediately •⁠ ⁠Start exploring alternatives NOW •⁠ ⁠Build something on the side •⁠ ⁠Answer: "What will you do when the gap closes for you?" •⁠ ⁠The day your salary matches a local hire's, your company will let you go •⁠ ⁠Plan for your 40s, 50s, 60s — don't just copy your parents' plan --- BOTTOM LINE: If you're an Indian IT employee earning well, you're on borrowed time. The system is designed so that once you become expensive, you're replaced. Plan accordingly.

A friend just sent me a Hindustan Times article. It was about me. A national newspaper wrote a full story on my post- the one where I said no to a 90% pay-cut "internship." No interview. Nobody reached out. I found out about my own news the way you find out about a stranger's. The surreal part: I'm 2 weeks from unemployment. A newspaper covered me refusing a job before I've found my next one. If you're hiring engineers, my DMs are open. Apparently I'm newsworthy now.

Fable is going to make a lot of people rich. Every model jump has minted a new batch of solo founders. GPT-4 gave us wrapper SaaS. Claude Code gave us one-person product teams. Fable just raised the ceiling again. I've shipped 3 products solo with AI in my stack. I know exactly what the last ceiling felt like. this time I'm not watching from the sidelines.

Good night everyone, have a wonderful tomorrow.

Man, today was a very exhausting day. I built a feature end to end - a very hard feature that will be the core of the product. It's really amazing. I worked with Mythos Fable - an awesome model, good with UI but freaking slow.

So I was using Cursor $60 plan and I used 1 billion tokens of Composer 2.5 non-fast mode and that's it. 60$ plan is gone. Very sad, man.

That viral moment - a Bollywood star glowing in a golden silk saree, baby bump on display at her new sea-facing home. Everyone's talking about that maternity fashion look. 🇮🇳✨ But here's the thing - what if a brand could launch an entire maternity fashion campaign inspired by that exact vibe... without a single photoshoot? No models. No studios. No 3-week production delays. Just type what you want, and AI generates photorealistic fashion campaigns from scratch. Luxe lehengas. Cozy co-ords. Any model, any setting, any aesthetic. We built it. It's called OLLI. olliforbrands.com Which Bollywood-inspired maternity look should we generate next - classic silk saree or modern co-ord set?

Composer 2.5 is really a good model. I really love working with it. For people who are using Chinese models, Chinese A models, I urge you to try Composer 2.5. It is cheap, it is fast, it is good, it has good design sense. I would say if you have the code base ready, if you have the structure of code base in a very good way, then it is easy for you to do the UI as well.

I calculated the real ROI of my degree 🎓 College: 🎓 St.Xaviers Fees paid: ₹1.0L Verdict: CRUSHED IT 📊 ROI on fees: 885× ⏱️ Payback: 1 months 🎯 ROI multiple: 885× 🍜 My fees = 6,667 Maggi packets ☕ A chai tapri breaks even faster than my college 💀 What's your college ROI? → paise-barbaad.netlify.app

Guys, we are facing a problem. Does anyone know how to stuff it? We are having S3 and 75% of the cost we are paying for S3 is for the movement of data - increase, outgress, ingress, egress, increase, egress. So, how to fix that? Anyone can tell me how to make S# cheaper?

Guys, if you want to get 5 million impressions in 3 months, then at least you need to get 60,000 impressions per day. 60k impressions per day. So at least you need to reply 200 replies a day to reach that. That is fucking hard work. Otherwise you need to be lucky.

Today I’m working with Hermes. We are creating a Hermes plug-in so that we can make Hermes our marketing agent – a guy who will sit in WhatsApp and help customers do everything. We are creating a plug-in and hope to open source it soon.

Guys, I am working on a very tough problem. It's about responsive testing. It is taking too much time for me to do responsive testing across all the different screens. I want to systemize it so my agents could easily do it. Do you think there is any better way? Have you guys faced that issue? Or do you think we have any solution for it? Feel free to comment down, guys. It's really helpful if you could give me some guidance on this.

So I am working on a UI heavy thing. I was using mostly GPT 5.5 for UI things. There was something that was not able to be fixed — GPT 5.5 was not able to fix it, so I lost my hope. I just threw it to Minimax M3. Man, it fixed it. Without sweating, it fixed it. I think there is something going on. Yeah, I can see the difference. Good going.

Today all my post is mini-max lol.