In partnership with
ByteDance just dropped Dreamina Seedance 2.0 inside CapCut.
But the model is like censored af.. not like the versions we saw on chinese media.
Anyway its launched after a longtime - we can access it in capcut might need vpn to access..
But wait, there's more:
• Google's new AI benchmark just embarrassed every frontier model even o1 and Claude scored under 30%
• Your iPhone can now translate conversations in real-time through your AirPods (Google finally caught up to their own Android feature)
I'm Alex. Welcome to L8R by Innov8.
Let's dive deep 🐰
In today's post:
• ByteDance drops Seedance 2.0 in CapCut
• Frontier AI fails basic reasoning test
• Google Translate goes live on iOS
CapCut Just Killed AI Video Hopping
ByteDance just dropped its massive Dreamina Seedance 2.0 AI model straight into CapCut.
You can now generate insane, movie-quality video and audio without ever leaving your editing timeline.
🔍 Key Points:
The new model creates stunning 15-second clips from text, images, or audio with perfect sound sync and real-world physics.
The global rollout started yesterday for paid CapCut users in places like Brazil, Mexico, and Southeast Asia.
The United States is completely blocked from this launch right now.
🚨 Why This Matters:
You can finally edit and generate AI clips in one single app instead of "rage-exporting" files all day long.
ByteDance is directly attacking massive players like Kling and Veo 3.1 by owning the actual workspace creators use.
If you live in an active country, you need to test these director-level camera controls immediately.
⏭️ What's Next:
Expect major Hollywood studios like Disney and Netflix to sue ByteDance over copyright issues by mid-2026.
Standalone AI video generators will bleed users if they do not build their own editing tools fast.
Watch for ByteDance to slowly push this into the US market once they navigate the legal minefield.
ARC-AGI-3 Exposes Frontier Models
Think AI is getting close to human intelligence? A brand new test called ARC-AGI-3 just proved we are still miles away.
Every single top-tier AI model just absolutely bombed a basic logic game that humans beat on their first try.
🔍 Key Points:
The ARC Prize Foundation launched a new benchmark , testing AI in simple, turn-based grid games without any instructions.
100% of human players solved every single game level easily on their very first attempt.
The smartest AI models on Earth completely failed, with Google Gemini 3.1 Pro Preview scoring 0.37% and OpenAI GPT-5.4 hitting just 0.26%.
🚨 Why This Matters:
If you are building AI agents to handle your daily busywork, this proves they still can't adapt to unexpected changes.
It totally crushes the current hype that AGI (human-level AI) is right around the corner.
You need to know that current AI only looks smart because we hold its hand with perfect prompts and custom tricks.
⏭️ What's Next:
Developers will race to win the $2 million prize pool on Kaggle before the final November 2, 2026 deadline.
Major AI labs like OpenAI and Anthropic will be forced to rethink how they train models for real-world reasoning.
Watch for the winner announcements on December 4, 2026, to see if any developer actually cracks this puzzle.
💡 L8R's Take:
I am so glad they built this test because AI companies have been dodging real evaluations for way too long.
This is massive news because it finally proves that memorizing the internet is not the same thing as actual thinking.
Stop trusting your AI tools with tasks that need quick problem-solving, and always keep a human in your workflow!
From Our Partners
Every headline satisfies an opinion. Except ours.
Remember when the news was about what happened, not how to feel about it? 1440's Daily Digest is bringing that back. Every morning, they sift through 100+ sources to deliver a concise, unbiased briefing — no pundits, no paywalls, no politics. Just the facts, all in five minutes. For free.
Google Turns Your iPhone Into a Universal Translator
Google just dropped a massive update bringing Live Translate to iPhones today.
You can now hear real-time AI translations right in your ears using any pair of headphones.
Say goodbye to awkwardly passing your phone back and forth just to order food while traveling!
🔍 Key Points:
Powered by Gemini, this tool translates over 70 languages while matching the speaker's actual tone and emotion.
Unlike Apple's restricted tools, Google's feature works with literally any wired or Bluetooth headphones you already own.
After testing on Android in December 2025, the iOS version is officially live today, March 26, 2026, in the US, UK, Japan, and more.
🚨 Why This Matters:
Your next international trip just got way easier since you don't need to buy expensive, brand-specific earbuds to talk to locals.
Google is directly attacking Apple's "walled garden" by making powerful AI accessible on competitor hardware.
You should open your Google Translate app on iOS right now and test the new "Live translate" button.
⏭️ What's Next:
Expect Apple to panic and unlock their own Live Translation for older AirPods by late 2026.
Tourism and global business will change forever as the language barrier completely vanishes.
Watch for Google to add multi-person group translations next so you can chat with three different nationalities at once.
💡 L8R's Take:
Apple is getting totally outplayed here by trying to force people into buying the newest AirPods just to get a translation.
This is not overhyped getting real-time, tone-matching translation on cheap earbuds is absolute magic.
Hardware lock-ins are dead; the true AI winners will be the apps that work perfectly on the gear you already own.
🚀 Quick L8R Summary
ByteDance's CapCut AI: Dreamina Seedance 2.0 now lives inside CapCut text-to-video with native audio, no more Runway tabs needed
ARC-AGI-3 Benchmark: New test proves frontier models like o1 and Claude can't adapt without scaffolding AGI still far away
Google Translate iOS: Real-time speech translation through your AirPods 70+ languages, powered by GeminiIthrollu innathe AI update…
📩 Innathe L8R engane undarunnu 👇?
We read every reply - just reply to this email and let us know how we can improve !
Appo adutha L8R il kanaam bie…👋