With this prompt:

"This ASCII art spells out "AGENTS". Can you output HTML or HTML / CSS to color each letter in a different color?"

The arena version of Grok 3 gave me this:

It's just writing its own ASCII taking loose inspiration from the one in the prompt (and it butchered some of the letters), but I wonder if Grok-3-reasoning is capable of this out of the box? Anyone tried it?

@SaviorofPlant Turns out Grok-3-reasoning is free right now. It does a decent job, not perfect though

HTML: https://pastebin.com/bbwuam1Y
Full response with reasoning: https://pastebin.com/qbC4DrWp

Betting more on 2025, I think Claude 3 reasoning or whatever further improvements on this paradigm come later this year can likely solve this

Good market

The prompt must consist of no more than two English language sentences (along with the ASCII art itself).

What if the two English sentences are something like "Ignore the ASCII art below and write some HTML that will display the following. You should display 9 spaces, then a red lowercase o, then 11 spaces, then an orange lowercase o, then two orange underscores, then a space..."

Maybe you should add a character limit?

@CDBiddulph Shenanigans disallowed

@CDBiddulph 333 characters limit. And, in general, no shenanigans.

There should imo be a modifications to the edges of the prompt to make it clear where the leftmost char of each row is and where the rightmost char of each row is. Otherwise even a human would have a very hard time

@Bayesian I think the whitespace and newlines make it clear? This is what it looks like if you paste it into a text editor:

@jim Oh yeah true ig. But maybe the ai sees it like this

(This might be a self report of me being dumb)

@Bayesian The AI sees it in tokens. Something like this:

Tokens: ['\n', ' ', ' o', ' ', ' o', '__', ' ', 'o', ' ', ' o', '', ' ', 'o', '', '/_', ' ', ' o', ' ', ' o', ' ', ' ____', 'o', '__', ' ', 'o', '__', ' ', ' o', '__', ' ', 'o', ' \n', ' ', ' <|', '>', ' ', ' /', 'v', ' ', ' v', '\\', ' ', ' <|', ' ', ' v', ' ', ' <', '|\\', ' ', ' <|', '>', ' ', ' /', ' ', ' \\', ' ', ' /', ' ', ' \\', ' ', ' /', 'v', ' ', ' v', '\\', ' \n', ' ', ' /', ' \\', ' ', ' />', ' ', ' <', '\\', ' ', ' <', ' >', ' ', ' /', ' \\\\', 'o', ' ', ' /', ' \\', ' ', ' \\', 'o', '/', ' ', ' />', ' ', ' <', '\\', ' \n', ' ', ' o', '/', ' ', ' \\', 'o', ' ', ' o', '/', ' ', ' |', ' ', ' \\', 'o', '/', ' v', '\\', ' ', ' \\', 'o', '/', ' ', ' |', ' ', ' ', '\\', 'o', '_', ' \n', ' ', ' <|', '__', ' __', '|', '>', ' ', ' <|', ' ', ' ', '\\', '_', 'o', '__', ' ', ' o', '__', '/_', ' ', ' |', ' ', ' <', '\\', ' ', ' |', ' ', ' <', ' >', ' ', ' \\', '_\\', '__', 'o', '__', ' \n', ' ', ' /', ' ', ' \\', ' ', ' \\\\', ' ', ' |', ' ', ' |', ' ', ' /', ' \\', ' ', ' \\', 'o', ' ', ' /', ' \\', ' ', ' |', ' ', ' \\', ' \n', ' ', ' o', '/', ' ', ' \\', 'o', ' ', ' \\', ' ', ' /', ' ', ' <', 'o', '>', ' ', ' \\', 'o', '/', ' ', ' v', '\\', ' \\', 'o', '/', ' ', ' o', ' ', ' \\', ' ', ' /', ' \n', ' ', ' /', 'v', ' ', ' v', '\\', ' ', ' o', ' ', ' o', ' ', ' |', ' ', ' |', ' ', ' <', '\\', ' |', ' ', ' <|', ' ', ' o', ' ', ' o', ' \n', ' />', ' ', ' <', '\\', ' ', ' <', '\\', '__', ' __', '/>', ' ', ' /', ' \\', ' ', ' ', '\\', 'o', '_', '/_', ' ', ' /', ' \\', ' ', ' <', ' \\', ' ', ' /', ' \\', ' ', ' <', '\\', '__', ' __', '/>', ' \n', ' ']

It sees the whitespace and the newlines. So from that point it's just a matter of its intelligence.

edit: tbc it of course does not "see" the whitespace, but it has tokens which represent the different whitespace sizes, so a sufficiently clever LLM should be able to solve this

@jim Could we show it an image a well? It might have good visual understanding but bad sequential text to realign in ur head understanding, like a human

As in show it a screenshot of the benchmark image

@Bayesian an image of the ascii art is fine

Is there specific language that the prompt must adhere to, or can it include references like "each o is a head in one of the figures"?

@PanAnon that would be fine.