Don't forget that GPT-6 Astra can do more than just use Blender 😅 Can also use the in-app browser to create architecture diagrams

diagrams in the browser look free the week still ends when the bucket does

Easier to generate it with gpt-image-2

The in-app browser piece is the more interesting capability here. Curious whether the diagrams held up once the architecture got non-trivial those tools tend to fall apart past a certain graph complexity.

Cost too many for this shit

the browser + diagram workflow is the bigger signal here. once models can move between research, tools, and visual output without hand-holding, “agent” starts feeling less like a product label.

Great wallpaper choice, which wallpaper did you use?

That's actually useful

the diagram only copies if it's a file. was Astra emitting mermaid or excalidraw, or clicking a generator in the browser and grabbing a png?

Is there anything Astra can't do?

The diagram can be the reply. Interesting consequences for what we call a page.

do you guys nerf the model a tiny bit after launch day? i can’t prove it but i feel it as a power user of codex/chatgpt

nice, how long that time take time to finish the diagram.

In light!!??

this way needs to spend too much token ~~~

@tldraw that you in the browser working with Astra?

The computer use of Astra is insane

That's funny... Everyone went hog wild with Blender. 🤣 I'm digging building full Powerpoint presentations without touching the keyboard. That's crazy.

browser diagrams are the quieter flex. do they stay editable after the session, or do you export a static image and lose the graph?

created a 3D org chart that looks like severance (friendly edition) hahaha but diagrams are so fun!