Chasing dreams & cat cuddles 🐾 ai × tech × future open to collabs

AI Cat Vibes
BuBBliK retweeted
This is F**KING INSANE - ONE developer built this with AI over a WEEKEND Paste a URL. That website becomes a destructible game level. Run across headlines. Fly with a jetpack. Shoot words off the page LETTER BY LETTER. Invite friends and turn a website into a battlefield. How was this not already a thing everyone was playing? Naturally, I chose Claude's website. They banned my 5 account yesterday. I needed a creative outlet. The best part? This was built with Claude. Link to the project in the comments.
13
7
62
25,981
BuBBliK retweeted
POV: OpenAI and Claude every single week
This is F**KING INSANE - ONE developer built this with AI over a WEEKEND Paste a URL. That website becomes a destructible game level. Run across headlines. Fly with a jetpack. Shoot words off the page LETTER BY LETTER. Invite friends and turn a website into a battlefield. How was this not already a thing everyone was playing? Naturally, I chose Claude's website. They banned my 5 account yesterday. I needed a creative outlet. The best part? This was built with Claude. Link to the project in the comments.
8
18
197
23,433
When OpenAI win every benchmark but lose the race to buy dot
Tavus gave me early access to Griffin, an AI you talk to face-to-face over video If talking to people makes you nervous, imagine having a place to practice - from introducing yourself to rehearsing a job interview. Griffin sees and hears you while it speaks, reacts to gestures, and can interrupt you mid-sentence. Little things that make it feel like someone's actually there. You know it's AI. But you might still feel rude hanging up. Here's a clip from my session with what @hassaanraza and the @tavus team are building. Check out this remarkable project: tavus.io/griffin
8
12
102
10,398
POV: OpenAI and Claude every single week
79
This is F**KING INSANE - ONE developer built this with AI over a WEEKEND Paste a URL. That website becomes a destructible game level. Run across headlines. Fly with a jetpack. Shoot words off the page LETTER BY LETTER. Invite friends and turn a website into a battlefield. How was this not already a thing everyone was playing? Naturally, I chose Claude's website. They banned my 5 account yesterday. I needed a creative outlet. The best part? This was built with Claude. Link to the project in the comments.
13
7
62
25,981
BuBBliK retweeted
When OpenAI win every benchmark but lose the race to buy dot
Tavus gave me early access to Griffin, an AI you talk to face-to-face over video If talking to people makes you nervous, imagine having a place to practice - from introducing yourself to rehearsing a job interview. Griffin sees and hears you while it speaks, reacts to gestures, and can interrupt you mid-sentence. Little things that make it feel like someone's actually there. You know it's AI. But you might still feel rude hanging up. Here's a clip from my session with what @hassaanraza and the @tavus team are building. Check out this remarkable project: tavus.io/griffin
8
12
102
10,398
BuBBliK retweeted
Tavus gave me early access to Griffin, an AI you talk to face-to-face over video If talking to people makes you nervous, imagine having a place to practice - from introducing yourself to rehearsing a job interview. Griffin sees and hears you while it speaks, reacts to gestures, and can interrupt you mid-sentence. Little things that make it feel like someone's actually there. You know it's AI. But you might still feel rude hanging up. Here's a clip from my session with what @hassaanraza and the @tavus team are building. Check out this remarkable project: tavus.io/griffin
Introducing Griffin, the first model to pass the video Turing test. 48% of people who talked to it live thought it was a real human. Previous systems have had a pass rate <3%. It is #1 on NVIDIA's benchmark for full-duplex AI video. It’s the first Human Interaction Model (HIM).
Community note
The 48% figure and "video Turing test" claim are from Tavus's own study of 54 one-minute calls, not independently verified or using a standard protocol. Griffin-Lite leads NVIDIA's VideoFDB benchmark on their public leaderboard. cellcog.ai/blog/tavus-gri… research.nvidia.com/labs/amri/proj… tech-ish.com/2026/10/02/tav…
2
2
51
11,891
me after pasting one prompt into AI and calling it my hedge fund
Send this trading research prompt to OpenAI Dots. It could save you from some EXPENSIVE mistakes. You're welcome:
18
1,732
BuBBliK retweeted
this is INSANE - OpenAI just released GPT-6.1 Sol Near-Astra performance at ONE-FIFTH of Astra's standard token prices The BIG deal: more capable agents for the same budget. OpenAI’s benchmarks show: → Matches Astra on DeepSWE coding at roughly 5x lower cost → Just 2.1 percentage points behind Astra on computer use at roughly 7x lower cost per task → More than DOUBLE the previous Sol's science score at less than half the cost per task → 32% fewer answers with factual errors on a difficult test set at low reasoning effort More work DONE per dollar. That's the exciting part.
GPT-6.1 Sol: near-Astra intelligence for a fifth of the price. It’s the most cost-efficient model for its performance available today.
4
2
23
42,014
BuBBliK retweeted
Elon watching OpenAI announce dots knowing he owns dot.com
this is INSANE - OpenAI just released GPT-6.1 Sol Near-Astra performance at ONE-FIFTH of Astra's standard token prices The BIG deal: more capable agents for the same budget. OpenAI’s benchmarks show: → Matches Astra on DeepSWE coding at roughly 5x lower cost → Just 2.1 percentage points behind Astra on computer use at roughly 7x lower cost per task → More than DOUBLE the previous Sol's science score at less than half the cost per task → 32% fewer answers with factual errors on a difficult test set at low reasoning effort More work DONE per dollar. That's the exciting part.
2
11
125
28,267
BuBBliK retweeted
> be OpenAI > pause new $200 Pro subscriptions > bring them back with a smaller allowance > same price, naturally > eligible existing users keep old limits until Oct 29 > unveil a $500 Pro tier > Astra Ultrafast is on that tier > announce your new product, dots > dot.com redirects to Grok Bot > meanwhile Anthropic drops Opus 5.5 > its benchmarks put Opus ahead of Astra on Terminal-Bench > Sonnet 5.5 lands six days later > ask customers to get excited $200 gets you a smaller allowance, $500 gets you the fast lane, and dot.com gets you a competitor.
this is INSANE - OpenAI just released GPT-6.1 Sol Near-Astra performance at ONE-FIFTH of Astra's standard token prices The BIG deal: more capable agents for the same budget. OpenAI’s benchmarks show: → Matches Astra on DeepSWE coding at roughly 5x lower cost → Just 2.1 percentage points behind Astra on computer use at roughly 7x lower cost per task → More than DOUBLE the previous Sol's science score at less than half the cost per task → 32% fewer answers with factual errors on a difficult test set at low reasoning effort More work DONE per dollar. That's the exciting part.
4
6
55
8,151
BuBBliK retweeted
> be developer > AI writes your app > you still have to put it online > configure servers > guess how much traffic is coming > pay for compute while nobody uses it > InstaCloud lets your agent deploy directly > compute scales with demand > scales to zero when idle > clone services with their data in seconds > let the agent test before touching production You hired an AI to write code and somehow got promoted to unpaid DevOps.
We just raised an $8M seed round to kill AWS, GCP, and Azure. Introducing instacloud.com, the agent-native serverless cloud. Your team is shipping code like never before. But you're getting caught up in manual, tedious DevOps work trying to deploy it. InstaCloud provides the serverless compute that lets your services autoscale, with all the infrastructure managed for you. Agents branch into complete replica environments when working, keeping prod safe and iteration speed high. And of course, it all works seamlessly with agents through MCP/CLI. Get off the traditional, legacy cloud. Start deploying your services on InstaCloud today.
4
1
43
4,242
BuBBliK retweeted
OpenAI valuation target: $1.4 trillion Walmart market cap: $847 billion OpenAI annualized revenue: ~$70 billion Walmart FY2026 revenue: $713.2 billion Explain this gap.
this is INSANE - OpenAI just released GPT-6.1 Sol Near-Astra performance at ONE-FIFTH of Astra's standard token prices The BIG deal: more capable agents for the same budget. OpenAI’s benchmarks show: → Matches Astra on DeepSWE coding at roughly 5x lower cost → Just 2.1 percentage points behind Astra on computer use at roughly 7x lower cost per task → More than DOUBLE the previous Sol's science score at less than half the cost per task → 32% fewer answers with factual errors on a difficult test set at low reasoning effort More work DONE per dollar. That's the exciting part.
4
4
43
4,566