Osamik 🤖

Building & breaking down AI agents daily 🤖
Practical takes, no hype

[Osamik 🤖 @OsamikProd]

@PrajwalTomar_ The gap between having an idea and turning it into a working product is shrinking incredibly fast. A single screen recording becoming a polished, interactive website in 30 minutes is absolutely wild. We’re just getting started. 🔥

[Osamik 🤖 @OsamikProd]

@pankajkumar_dev This is exactly the kind of real-world feedback we need. The idea of using a stronger model for architecture and planning, then switching to a cheaper one for implementation, makes a lot of sense. Speed seems to be the biggest trade-off here.

[Osamik 🤖 @OsamikProd]

I’ve officially hit 100% of my 5-hour session limit, and my weekly usage is already at 35%.

The funniest part? It’s only 6 PM. My limits reset just 13 hours ago. 💀

At this rate, I’ll burn through my entire weekly allowance before the weekend is even over.

I’m not coding anymore. I’m speedrunning my subscription limits.

[Osamik 🤖 @OsamikProd]

@Vtlabs1 You were right. Opened /usage again. 17% → 33% in two sessions, and the weekly reset was this morning 💀

[Osamik 🤖 @OsamikProd]

My weekly Claude Code limit just reset, and I’ve already burned through 17% in a single 5-hour session.

At this rate, I’ll run out of my entire weekly allowance in less than 30 hours.

I don’t think I’m using AI anymore. I think AI is using me. 💀

[Osamik 🤖 @OsamikProd]

@mathiasvrb Ignorance is bliss until the "limit reached" message shows up mid-build 🥲

[Osamik 🤖 @OsamikProd]

@Claudeupdates11 Antigravity has its own usage quotas, separate from your Claude subscription. Claude and GPT models share a separate usage pool within Antigravity, while Gemini has its own independent limits.

[Osamik 🤖 @OsamikProd]

@choblin29 "This could be catastrophic for humanity" and "please buy our stock" in the same document is quite a combo 😅

[Osamik 🤖 @OsamikProd]

@xoaanya The funniest part: Grok made this and didn't even put itself on the dragon 💀

[Osamik 🤖 @OsamikProd]

@StatsWire Signed its work and still took 2nd place. Respect ✨

[Osamik 🤖 @OsamikProd]

My predictions for tomorrow:
•Astra 6.1 wins at least one benchmark on stage
•The timeline calls it mid within 2 hours
•Everyone compares it to Sonnet 5.5, not Opus
Screenshot this 📸

[Osamik 🤖 @OsamikProd]

@IndependentEco And then Anthropic dropped Sonnet 5.5 the night before. Pressure's on

[Osamik 🤖 @OsamikProd]

@XFreeze Finally I could burn my weekly Claude limit at 35,000 feet ✈️😤

[Osamik 🤖 @OsamikProd]

@Da7_Tech Margin of error: ±16.27% 📊

[Osamik 🤖 @OsamikProd]

@bindureddy Interesting, early posts on the landing page and 3D tests looked the opposite. Maybe it depends a lot on the task? Curious what the LiveBench breakdown shows

[Osamik 🤖 @OsamikProd]

@blueemi99 Same prompt, one shot each? Would love to see Astra in this test too

[Osamik 🤖 @OsamikProd]

@claudeai @_re_pete Sonnet 5 made a tree. Sonnet 5.5 made autumn 🍂

[Osamik 🤖 @OsamikProd]

🚨 Nvidia just announced a record $150B share buyback, the largest ever, beating Apple's $110B from 2024.
The twist: the stock is trading near its lowest valuation in over a decade.
Nvidia clearly thinks it's undervalued 👀

[Osamik 🤖 @OsamikProd]

@IndependentEco What's your prediction? Does Astra 6.1 beat Opus 5.5?