It's OVER.
GPT Image 2 is out! It's insane at text now. Posters, labels, typography, all clean. Still somehow loses a street fight against counting the Rs in strawberry. It’s so OVER! Great at info graphics and small texts and thumbnails too. is it available to pro users? GPT Image cooks logos, signs, and clean typography now. Then strawberry shows up and the whole illusion collapses. It’s OVER.

Nano Banana killer is here? Exciting news - GPT-Image-2 by
@OpenAI
has claimed the #1 spot across all Image Arena leaderboards!

It’s OVER.
I vibe coded a trolley dilemma game in
AnthropicAI
Opus 4.7 (oneshot, 30 levels!) where every choice gets more cursed.
Kitten vs AI chips.
Robots vs humans.
Logic vs chaos.
Play here https://vortex.channel/trolley/

Smart Shot on Openart is wild.

You type one line, like a billionaire giving a ridiculous Mars mansion tour with Tesla robots as servants and a robot dog. Or Tokyo style streetware ad...

It builds a full GPT Image 2 Shot Plan Sheet first, characters, set, storyboard, camera, lighting, lenses.

Then Seedance 2.0 shoots it.

Not random AI video anymore, this feels directed.

Try it on https://openart.ai/

A new Qwen-Image-2.0-Pro has been released.

Over the past year, descriptions of new features for any generator all read like copy-paste: improved prompt adherence, better text rendering, improved consistency. Truly new features—like layers—rarely appear. So the main metric ends up being “by eye.”

And by eye, the new Qwen ranks 9th on the Arena (and 17th in Edit (!)).

And for now, the most disappointing part is that it isn’t being open-sourced (and from what I’ve read, it may not be at all).

Dubbing with lip sync

Check out this two-minute video to see how lip-synced dubbing from sync.so works now.

The new dubbed language now matches the speaker’s lip movements.

In principle, sync.so could already adjust lip movements to fit a given audio track, but now they’ve added translation and voice cloning. It’s basically turned into a full lip-synced dubbing studio.

It’s interesting that the example is shown in a medium shot—we’ll have to see how it performs on close-ups.

As always, the pricing is steep. But it looks like the best lip-sync tool on the market right now.

[https://sync.so/pricing]

For geeks: LTX2.3 VR-Outpaint IC-LoRA
This is a 360° video outpainting LoRA — it takes a regular video as input and outputs a VR-ready equirectangular video.
LoRA and workflow here: https://huggingface.co/TheBurgstall/VR-360-Outpaint-LTX2.3-IC-LoRA
Comfy node: https://github.com/Burgstall-labs/ComfyUI-EquirectProjector

Grok 4.3 has been released.
And of course, benchmarks immediately appeared where it outperforms everyone else.
It’s already available on OpenRouter: [https://openrouter.ai/x-ai/grok-4.3]

Pricing: $1.25 per 1M input tokens and $2.50 per 1M output tokens.
But the most interesting part is Grok Imagine Agents...

ElevenMusic isn’t just a music generator.

It’s not a competitor to Suno, but rather ElevenLabs’ attempt to build a full music platform with AI generation, remixing, a discovery feed, and payouts for creators.

The user doesn’t just listen to a track — they can:
listen to independent artists → remix their tracks → create their own versions using AI → publish the result → earn money if the music gets streams.

ElevenLabs explicitly describes ElevenMusic as a platform that combines discovery, remixing, and original creation in one system. In their wording, this is meant to give artists a “direct path to monetization,” and fans the ability to participate in music rather than just consume it.

It’s something like an AI-powered Spotify + TikTok-style remix culture + a marketplace for music licensing.

The question is what happens first: will Spotify get flooded with AI-generated music, or will ElevenMusic manage to carve out a share of that market?

https://elevenmusic.io/

Neural rendering in Seedance

This is really good. Previs and animation in Blender, neural rendering in Seedance.
Instead of texturing — reference images.

Take a look here for more:
https://x.com/fatboypink

This AI Agent Can Run Your Entire Business While You Sleep

Accio Work is not just another AI chat app. It is a 24/7 agentic business team built to actually execute tasks, not just answer questions.

It can research markets, monitor competitors, source products, negotiate, launch stores, promote on social media, manage operations, organize files, connect to tools like Gmail, Google Drive, Slack and Notion, and even automate scheduled workflows.

With Agent Hub, you can create custom AI agents, choose between models like Gemini, GPT-4o, Claude, and Qwen, give them skills, connect them to channels like Telegram or Discord, and let multiple agents work together inside one workspace.

The wild part? Accio Work is local-first, permission-based, sandboxed, and built with enterprise-grade security, so you stay in control of what your agents can access and do.
👇👇
https://www.accio.com/work?src=p_ttkol_vortexnextgen

Another leak from the new Gemini Omni Model

Here’s proof that this was actually made in Gemini:
https://gemini.google.com/share/7d5dc678c80a

Prompt: A professor writes out a mathematical proof for trigonometric identities on a traditional chalkboard, explaining the step he is currently on in the equation.