Hey, it's the Carat team. 🤗
Today: Claude Opus 4.7 is out, AI takes direct control of your Mac, and turning a single photo into a 3D world.

Anthropic just released its new model, Claude Opus 4.7. They announced it would come this week just yesterday, and it's already here.

The biggest change is that it handles long tasks thoroughly from start to finish. If it makes a mistake, it self-checks and corrects. Coding performance jumped significantly (7-11% on industry benchmarks), and image understanding tripled. The price stays the same as the previous model.

What's even more notable is the design tool that launched alongside the model. You can now create websites and presentations right inside Claude. This fits right in with the app builder we covered two days ago.

OpenAI's Codex got a massive update. AI now opens Mac apps, clicks, and types on its own. Watch the video and you'll see the cursor moving by itself.

Image generation (gpt-image-1.5), 90+ external tool connections, memory that remembers past actions, and task automation. It's no longer just a chatbot. It's becoming an AI that directly uses your computer.

NVIDIA unveiled Lyra 2.0. Feed in a single image and it transforms into a 3D world you can explore.

It outputs in 3D Gaussian, mesh, and physics engine formats, making it useful for games, architecture, VR, and more. The era of creating 3D content from a single photo is here.

Boston Dynamics' four-legged robot Spot now runs on Google's Gemini Robotics. It understands its surroundings, identifies objects, and follows simple commands.

Tell it to "clean the room" and it picks things up off the floor and puts them in a box. Following yesterday's Toyota basketball robot, AI robots handling everyday tasks is getting closer by the day.

Upload a selfie and AI imagines what happened to this person 5 minutes before the photo was taken, then generates it as an image.

For example, upload a smiling selfie and you might get: "5 minutes ago she was crying on a bench after getting dumped, then instantly pulled herself together." Same outfit, completely different expression and setting.

Generate the scene from exactly 5 minutes before this captured moment. Show the person in the same outfit but in a different emotional state and environment that narratively explains how they arrived at this current scene.

Try it on Carat's Nano Banana with your own selfie. Send it to a friend for a fun reaction.

That's all for today. We'll be back tomorrow with more useful updates. ☺️