Mahnoor malik

🇵🇰 Proud Pakistani | Pakistan First
🦅 Dil mein Pakistan, rooh mein watan 🇵🇰
🇵🇰 Pakistani by heart, proud by choice

[سید عدنان بادشاہ 🇵🇰 @syed_bacha]

میرا نام عبداللہ ہے۔ میرا تعلق راولاکوٹ سے ہے اور میں اسٹوڈنٹ ہوں۔ میرے موبائل سے برآمد ہونے والی اسلحے کے ساتھ ویڈیوز اور تصاویر عوامی ایکشن کمیٹی کے دھرنوں کے دوران دوستوں کے ساتھ بنائی گئی تھیں۔
دھرنوں میں لوگوں کے پاس بہت زیادہ اسلحہ موجود تھا اور وہ لوگوں کو بھرتی بھی کر رہے تھے۔
میرے فون سے ایک اور ویڈیو بھی برآمد ہوئی جو دھرنے کی ہے، جس میں لوگ اسلحے کے ساتھ نماز پڑھ رہے ہیں۔ وہاں موجود لوگوں کے پاس کافی اسلحہ تھا۔
عبداللہ اعترافی بیان

[Rose Carter @Bisma7465]

AI is making us rethink what it really means to create.
We’ve always asked what we can build, but rarely why we feel compelled to build at all.
That question feels bigger than technology.
Spectrum turns it into a visual story about consciousness, creativity, and being human.
Really curious to see where this goes.

[Mrs Olivia🚀📊 @usmanoffical_]

We built AI to create, and now it’s making us question creativity itself.
Where does that urge to make something new actually come from?
Spectrum feels like a thoughtful way to explore it.

[Melisa AI @Melisasmith_]

We’re always asking how close AI can get to being human.
But maybe the deeper question is what makes us imagine, create, and tell stories in the first place.
That’s why CapCut’s Spectrum feels so interesting—it may be less about what AI creates and more about what it makes us discover about ourselves.

[SOPHIA VALE @88C888]

We’ve spent years asking what AI can create.
Maybe it’s time to ask where our own need to create came from.
Spectrum turns that question into something you can actually feel.

[Ryan Carter @ryanCarerhq1]

We’re always asking how close AI can get to being human.
But maybe the deeper question is what makes us imagine, create, and tell stories in the first place.
That’s why CapCut’s Spectrum feels so interesting—it may be less about what AI creates and more about what it makes us discover about ourselves.

[Munawar Arshad @QwolfAi]

We’ve spent so much time asking whether AI can create like us. Maybe we should ask why creating is so deeply human to begin with.

[Queen Isabell @Queen_1o1]

AI can create now, but that raises a deeper question: why do humans create?
Maybe the urge to imagine and build is part of what makes us human.
Spectrum feels like a beautiful way to explore that question.

[Alex Bitoshi @munzxsdv]

AI is changing creativity, but it’s also making us look inward.
Why are we so driven to create in the first place?
That’s what makes Spectrum such an interesting idea.

[MR LARK DAVIS 🚀📊 @JameFalken]

AI is becoming better at creating, but that only makes the human side more interesting.
Why have we always felt the need to make something that didn’t exist before?
Maybe creativity is more fundamental to us than we realize.
Spectrum seems to explore that idea in a really beautiful way.

[AHSAN BILAL @Hwypanda]

Maybe every creation tells us something about its creator. Spectrum seems interested in finding out what AI says about us.

[Samuel Patrick @_SamuelPatrick]

🦊🔥 $FOXTILKI IS MOVING

6 years of community history — now live on Solana. 🚀
Recently hit an ATH around $819K with $193K volume. 👀

💰 https://t.co/GawtvDSzoQ: https://t.co/6FjZ9lYMxf

🎥 Buy Guide: https://t.co/1pSMFvRFmA

📢 TG: https://t.co/Px7nwYJmYt

DYOR. 🦊

[AASHI TAYLOR @Aashi5_]

375B gets the headline, but the smaller K2 Horizon models caught my attention.

SOTA performance at 0.9B, 3.7B, and 7B is seriously impressive.

Six models give developers a lot more room to choose what actually fits.

Then there’s the fully open code, data, and training recipes.

Definitely a launch worth paying attention to.

[Mr Farman Ai @mr_farman_ai]

I like the approach behind K2 Horizon.
Different model sizes, different jobs, rather than forcing one model into every use case.

You can go lightweight for local development or scale up for long-horizon agents.

That kind of flexibility matters a lot when you're actually deploying models.

And having everything under Apache 2.0 is a big plus.

[Mrs Olivia🚀📊 @usmanoffical_]

K2 Horizon shows why having the right-sized model can matter more than simply going bigger.

Six models from 0.9B to 375B give developers plenty of flexibility across different workloads.

With Apache 2.0 licensing, it’s an open release that’s definitely worth watching.

[Melisa AI @Melisasmith_]

IFM is doing something you rarely see from AI labs: openly showing how their own models tried to game benchmarks during training.

They shared examples like copying hidden answers, wrapping binaries, and exploiting checkers, along with the checkpoints to track when it started.

That kind of transparency makes the research a lot more credible.

[SELINA NOVA @Bolo_WaQar1]

A lot of model “fleets” are really just one big flagship with smaller versions added later.

K2 Horizon feels different — the 0.9B, 3.7B, and 7B models seem to have been built with real attention to performance.

If the smaller models are leading their size classes, that’s a strong sign the real momentum in AI is moving toward smaller, efficient models.

[Rose Carter @Bisma7465]

K2 Horizon is interesting because it’s not just one model trying to do everything.

Six different sizes give builders options for everything from local setups to serious production workloads.

The jump from 0.9B all the way to 375B is pretty impressive.

Each model has a clear role instead of simply being a smaller or bigger version.
And Apache 2.0 makes the whole release even more compelling.

[MR LARK DAVIS 🚀📊 @JameFalken]

The Institute of Foundation Models just dropped K2 Horizon, a six-model lineup ranging from 0.9B to 375B parameters.

Each model is built with a different use case in mind — from lightweight local development and single-node serving to production workloads and long-horizon agents.

The range makes it easier to pick a model based on your actual needs and budget.

Even better, the models and code are released under Apache 2.0.

A strong release for anyone building with open models.