Sad that “crafted by hand” immediately makes me think Claude made it.

> Four variants, one axis

Dunno about the font, but the copy on this page is 100% Claude.

It obviously wrote the article so why not.

Late to this party, but wanted to share my favorite piece of writing about how annoying AI-generated prose it to read: https://storage.courtlistener.com/recap/gov.uscourts.med.612....

I used to work regularly on a commuter train with a firm-provided touchscreen ThinkPad. The touchscreen was great on the train because it was faster (for many things) than reaching for the trackpad. And when you're in a sort of cramped sitting position, reaching forward feels natural.

I'm a lawyer by trade and am forced to lean on GUIs with little keyboard-control support. When I code (for pleasure), I don't miss the touchscreen, but I often miss it when I'm trying to read PDFs and Word documents on the go.

It would be interesting to know what ThinkPad model it was. The older ones had pretty small trackpads so they were difficult to reach.

On Macbooks the trackpad is so large that you can reach it even from the sides of the laptop and the two finger scrolling actually works without second thought.

I had the Thinkpad T480 with a touchscreen. Thinkpads were really awesome, you could easily fix that thing by buying parts from eBay.

Probably a yoga by the sounds of it

You got me curious, which GUIs with little keyboard-control support are you using / forced to use?

I'm assuming this is a post-train of Argon ("Building on Gemini 4 Argon, which was launched two weeks ago, Carbon illustrates the pace at which we’re pushing the frontier forward."), but does anyone know for sure? The post isn't clear.

the rumour mill the last two weeks has been referring to carbon and argon as "checkpoints" of the same model

That's generally how models are made. There's several candidate checkpoints and experimentation going on.

are we that much in a rush that we need to release checkpoints this close? or maybe argon was a little rushed?

I assume weights can also be IP in the form of trade secrets.

Wouldn't that be the processing of obtaining the weights, and not the weights themselves?

Based on how vibe-coded this whole thing is, I’m guessing the newer models weren’t in the training data and the agent just picked what it knew existed.

There’s a whole niche of legal scholarship around this idea (and some startups): https://papers.ssrn.com/sol3/papers.cfm?abstract_id=5331811

Kieffaber spoke at my law school, dazzling and mildly concerning (like, "is he on something?" concerning) speaker

Pricing is available here: https://dev.meta.ai/models/muse-wonder

Input -- $4.00

Cached input -- $0.20

Output -- $20.00

Edit: Upon closer inspection, the benchmark scores for "Muse Wonder" on the main marketing page are actually for "Muse Wonder Full," which is only available on the higher-tier subscription plans. This pricing is for "Muse Wonder."

200% Sol 6.1 pricing for Luna performance.

> Luna performance

Source? On the available benchmarks it looks like Sol 6.1 level.

sol 6.1 is 50% cheaper than this. sol pricing$2.00 output $10.00

dont know why the f people downvote for saying sol pricing.

It sounds like you're disagreeing with the person you're responding to, but you're both saying the same thing.

i was agreeing with that.

> Zuck is proven once again as a great executioner

Executor is the word, but everyone also knows what this was supposed to mean.

Let’s be honest it’s funnier as written.

Or Lecun was just bad.

couldn't feel the AGI

If computers can’t make decisions, can humans? Relevant: https://plato.stanford.edu/entries/computational-mind/

edit: This wasn’t meant to be a rhetorical question!