I dreamt about this.

And then I woke up and made it happen.

A is Kling 2.5. Yup. Just came out this week.
And B is Veo 3 (the so-called champion).

I’ll take you through different prompts i gave, and the output 👀

Spoiler: the above results are nothing, wait till you see the other outputs 😉

But before that, here’s what happened this week:

1. ChatGPT Pulse: Proactive Daily AI Briefings

OpenAI launched ChatGPT Pulse for Pro users, delivering personalized morning updates based on your chats, calendar, and connected apps. The AI researches topics overnight and presents 5-10 visual cards with news, reminders, and tailored insights. Users can curate topics and provide feedback to improve relevance over time.

2. Google Mixboard: AI-Powered Mood Board Creator

Google Labs released Mixboard, an AI tool that turns text prompts into visual mood boards using Gemini models and "Nano Banana" image editing. Users can create boards for home decor, events, or product ideas, then edit with natural language commands. Available in beta for US users only.

3. SeedVR2: One-Step AI Video / Photo Restoration

ByteDance released SeedVR2, an AI model that restores low-quality videos to HD in a single step using diffusion adversarial training. It handles any resolution and processes locally on NVIDIA GPUs with 8GB+ VRAM. The tool competes with Topaz Video AI for professional video enhancement.

Okay, you’re still here. Good. Because this is where it gets interesting.

I decided to put both models through the ultimate test: not just simple prompts, but cinematic, complex scenarios that would break most AI video generators (it did break one of them).

The Setup: Three elaborate prompts. Same input image.
Two titans of AI video generation going head-to-head. (here are the other two)

Let me show you what happened...

Test 1: Futuristic Samurai Duel

The Prompt: In the scene, the provided image of [USER] stands in a futuristic arena, dressed in read full prompt

My opinion: Both weren’t perfect. A mask started appearing on my face in kling, but in veo the physics broke instantly. BUT

The beauty of the movement is visible in kling. And that’s what it stands out for.

Test 2: Dance Battle Performance

The Prompt: In the scene, the provided image of [USER] is the central performer on a brightly read full prompt

My opinion: Veo3 looked freaking beautiful. But it was just moving my head. Not following the prompt at all. Kling on the other hand literally made into a popular south indian dancer.

What This Means

Remember when Veo 3 dropped and everyone called it the "GPT-4 moment" for AI video?

Well, Kling 2.5 just had its "Claude moment"  coming from behind to absolutely dominate.

WAN 2.5 also launched but the output queue was so long that it’ll be covered in an upcoming issue 👀

TLDR: Kling 2.5 handles complex human movement better than anything I've seen. The body tracking, the facial consistency, the way it maintains character integrity through impossible action sequences, it's genuinely scary good. At the rate we’re going im not sure what pre-production, production, and post production teams will do moving forward (seems like n8n flows will be there to just automate it end to end) 😢

At this point, it isn't about which tool wins anymore. It's how fast you can adapt when the benchmarks change every week.

Until next time,
Vaibhav 🤝

If you read till here, you might find this interesting

#AD 1

Clearing inboxes won’t grow your business. Wing gives you a dedicated virtual assistant to run scheduling, admin, and follow-ups. So every hour you save can turn into momentum you can scale.

Reclaim your hours now

#AD 2

Typeless turns your raw, unfiltered voice into beautifully polished writing - in real time.

It works like magic, feels like cheating, and allows your thoughts to flow more freely than ever before.

With Typeless, you become more creative. More inspired. And more in-tune with your own ideas.

Your voice is your strength. Typeless turns it into a superpower.

Download for Mac today