That's miles above what most other libraries and tools provide, seems great to me.

This is the clean grid. Maybe an additional test worth trying is to have Nano Banana Pro try to search and highlight just one specific word at a time, both without the solution in the prompt and with the solution in the prompt.

Image 1: directly asking NBP to find each word in the grid.
Image 2: asking NPB to highlight each word in the grid, with the position of every word included in the prompt.
Image 3: asking Gemini 3 to iterate on Image 2 and the prompt to see if it would make a difference.

I think the next step-up in capabilities will be when image models are also able to one-shot problems. Your post made me curious if giving Nano Banana Pro the answer in the prompt would help, so I had Gemini 3 perform the word search and prepare a prompt I could pass to NBP.

Lots of problems have moved really quickly from "able to be solved with some nudging / prompting" to "can be one-shot without special prompting". I had a simple word search problem that this year went from complete failure, to possible with nudging, to one-shot.

Thank you for your service o7

This is the bright red line. If any president tries to run for a third term without a constitutional amendment, their removal through the use of force is justified.

That cult pin goes hard, thank you for sharing.

With that being said, I do appreciate the outreach and commitment to engage with the community in order to gather feedback! Thank you for your work and efforts.

In particular, I don't even understand the kinds of use-cases where some of these CSS features are desired (not that I disagree, I'm just confused).
While other standards like e.g. Federated Credential Management or OpenID4VC, are things I haven't even heard of until now.

A bit of constructive feedback: some of these proposal descriptions / titles are very barebones and confusing to understand unless you dig in pretty deep to figure out what is meant. It would be helpful to have a little bit more context to go from.

Personally, I only indexed "what direction is Left", while Right is encoded as "Not Left".

What do you have in your cognitive fast index?

Let me give an basic example: the ability to instantly know Left or Right. For many people who I questioned, this didn't get added to their fast index until they start driving, when distinguishing directions instantly becomes needed.

absolutely goated, thanks to everyone!!!