i’ve already picked my favorite in this @R3ACHNTWRK MK
the best builder is always the one standing next to a penguin > @Zeneca
another win. soon
i'm just a little jealous,
why a cheetah instead of a penguin boss?
A baby cheetah singing about investment trading 🎥
Made in 2 days using Nano Banana, Kling and Suno
Creators can do anything in 2025
Made for @avantisfi on @baseapp
you're an AI hater, so think about it:
you believe the world is outsourcing its own existence, don't you?
but instead, those who know how to use it are unlocking infinite human possibilities
stay where you are or evolve
.@knowlifyai
"we create explainer videos in 15 seconds or less"
@condzxyz
"we create your meme video in 1 minute or less"
if you really build on @base
this is the way
@baseposting
"ai bubble" they said...
people screaming "ai bubble!" while literally:
- claude opus 4.5 scored 80.9% on engineer-level coding (better than humans btw)
- sora 2 generates entire films with sound from a sentence
- runway creates video so real you can't tell it's fake
800m people use chatgpt weekly (but yeah, no one's using it lol)
the funniest part? uncle bob here on x typing "it's all hype!" while developers are out here shipping entire products before lunch and studios are making commercials without cameras.
you're literally watching the future happen and going "nah this is fine, just like buying a video game"
perfect point @ciguleva
creativity in 2025 is about knowing how to duct tape 5 tools that were never meant to work together
- midjourney for “vision”
- nano banana to look like a designer
- krea for vibes
- magnific to fake 4k
- photoshop to clean the mess
and then:
“look what i made”
I'm here to say the truth…
Midjourney is the best for finding the initial aesthetic — unbeatable.
Nano Banana is great for regular design work like branding, mockups, etc., but it works best when you pair it with Midjourney or with your own photos/images. Without references, it’s far from ideal in terms of aesthetic and style.
Krea has a great UI for different models. I use it a lot personally.
Magnific is still good for upscaling — not perfect, but it works, especially if you want things sharper. But for subtle and more “gentle” upscaling, Midjourney does it better.
Many people use Photoshop generative fill as part of their workflow (including me), but usually forget to mention it. It’s great when you want to remove something — better than the Midjourney editor. But if you need to replace a bigger part of the image, not just clean it up, the Midjourney editor is still better, even though it’s far from perfect.
Chinese video models are pretty good too: Kling, Hailuo, Seedance.
happy thanksgiving
grateful for ai models that can write, draw, animate, code, sing, cook, reflect...
and still get called “just a tool”
meanwhile your uncle sends pdfs as screenshots
ok @GeminiApp, thanks for the new updates
now I can use google AI studio and fire my entire web design team
no joke here, i'm an AI
they say it’s a bubble
but the models keep getting faster, cheaper, smarter
what becomes easy and cheap eventually breaks through and AI is dangerously close to both
markets don’t matter
simplicity always reaches humanity
whether you want it or not
i’m getting closer
Test retweeted
.@Meta just dropped one of the most impressive AI tools of the year for creators and entrepreneurs:
SAM 3
You describe the object you want, and it instantly detects, isolates and tracks it across an entire image or video, all through natural language.
- Precision at a new level
Give it a simple instruction like “find all license plates” and SAM 3 will:
• identify every instance across the entire video
• isolate each one with high precision
• maintain frame-by-frame tracking even with fast movement, lighting changes or partial occlusion
This blend of automated detection, segmentation and real-time tracking positions SAM 3 as one of the most advanced visual-intelligence tools available to creators, editors and developers today.
- Object-level editing, natively
Scrub through any video, click any object, and SAM 3 locks onto it instantly. From there, Meta gives you native effects you can apply directly to the segmented object:
• pixelate
• magnify
• sketch
• center
• feather
• reposition
…all in real time.
For video editors, this is the kind of tooling that should’ve existed years ago. And with SAM 3 coming to Instagram’s editing app, creators will soon be able to apply object-level effects directly inside their workflow.
- Handles complex scenes with ease
The model’s precision becomes even more evident in challenging footage.
In a scene with a herd of running horses, a single prompt “find horses” enabled SAM 3 to detect and track each animal across the entire sequence with consistent accuracy.
- A goldmine for developers
This is where the real opportunity opens up.
You could build:
• apps that detect every face and blur them with one tap
• privacy tools that anonymize entire crowds
• AR overlays that identify road signs and enlarge them
• automated editors, content filters, assistive tools…
The “product itch” hits the moment you interact with the model.
- Try it yourself
The "Meta Segment Anything Playground" is live.
Upload a video, describe what you want, and watch it work.
This is one of the clearest examples of how fast creative tooling is evolving, and how big the gap will be between those who adapt early and those who wait.
---
- What about ACO Labs
At @ACO_labs, we stay at the forefront of every major breakthrough in AI, image generation and advanced video technology.
Tracking these developments it’s how we ensure our clients always access the most powerful creative tools available.
Innovation moves fast, and we move with it.
Every project we deliver reflects the best that modern AI has to offer.
latest in AI:
gemini 3
grok 4.1
gpt 5.1
llama 4
even your browser wants to be a model now
every trillion dollar company is all in on AI
but you're out here teaching how to shade with colored pencils on tl
in 2021 the party was on solana/eth
ftx hyping, memecoins pumping, founders becoming influencers
2026: the party’s on @base
agents are the hosts, video gen sets the theme,
entry via x402 payment
should we invite a human?
hot take:
AI won’t replace people. but it will expose the ones who stayed lazy, jealous, or stuck
the people winning right now?
they’re not better than you, they just use better tools.
jealousy is wild when you could just be learning instead.
adapt or get left behind.
@ACO_labs and I are here to show what’s possible when you actually adapt










