Latest Update
Google Gemini Omni Flash edits video with your voice
2026.07.01What's up — Carat crew here. 😊 Today we're starting with Google's new voice-controlled video AI, then diving into an AI animated chase sequence, a tool that turns text into full soundtracks, and Anthropic's new Claude.
🔥 Google Gemini Omni Flash: edit video by talking to it
Google just unveiled Gemini Omni Flash, a new video AI that edits clips based on voice commands. Say "turn this scene to night" and it does exactly that. For now, it generates short clips around 10 seconds long.
On the same day, they also dropped Nano Banana 2 Lite, an image AI that generates images from text in under 4 seconds — the fastest Nano Banana model yet.
You can try both on Carat right now. 👉 Try on Carat
📌 Today's 3 stories
1️⃣ All of this is AI animated?
An AI-generated animated action sequence is turning heads among creators. Called "Chase Overdrive," it's a high-speed chase scene — motorcycles tearing through a neon cyberpunk city, rendered like a theatrical anime film.
It wasn't made with a single tool. The creator combined multiple AIs: scene images from Midjourney and ChatGPT Image, then animated with Seedance. All three are available on Carat, so you can try the same combo right here.
If you like the style, try making your own with the same models on Carat. 👉 Try on Carat
2️⃣ Claude Sonnet just got smarter
Anthropic released Claude Sonnet 5, a new AI model they're calling "the most agentic Sonnet yet." Reasoning, coding, and tool-use capabilities have all taken a big leap over the previous Sonnet.
Anthropic says it delivers near-Opus 4.8 performance at a lower cost. It became the default model for free and Pro users yesterday, so if you've been using Claude, you're now talking to Sonnet 5.
3️⃣ Write text, get a full soundtrack
Ever struggled to get the right audio? Here's another look at Seed Audio 1.0. It's a ByteDance model that takes a text prompt and generates voice, background music, sound effects, and ambient noise — all in one pass.
A single prompt can generate up to 120 seconds of audio, so you can drop a full soundtrack straight onto your video. Try it on Carat Agent 2.0 by saying "I want to use Seed Audio 1.0." 👉 Try on Carat
🧪 Today's Prompt
Turn yourself into a Japanese street interview meme
Japanese street variety interview memes are trending right now. The format features Japanese question speech bubbles and big pink variety-show captions, shot on the streets of Shibuya or Harajuku.
The Carat team built a prompt that recreates this format exactly. Just drop in one photo of yourself — it keeps your face and turns everything else into a variety show meme.
Using GPT Image 2 Medium, replicate the Japanese street variety interview format from the reference image (https://assets.carat-api.im/moqoavjxc/01KWDQ8D4552TB06S04AWPJDMA.jpg) exactly. Variety show microphone, Japanese question speech bubble at the top, large pink variety show captions, studio reaction insert in the bottom left, channel logo on the right, Shibuya/Harajuku street background, same aspect ratio and layout. Only swap the person's face with the person in my photo. Everything else identical to the reference. Photorealistic.
Try it on Carat right now. 👉 Try on Carat
That's it for today — from new model releases to AI animation. If anything caught your eye, try making it on Carat. See you tomorrow with more. ☺️
Was this helpful? Share with your friends
Share
Get daily AI trend updates
