simlevesque

They pionneered unified memory a few years ago.

You buy an Amazon gift card as a gift and use the email Anna's Archive provide you as the recipient. The email is unique to you so they know when you paid them.

I was blown away when I paid to get an API key for it. The process is intricate and state of the art.

It doesn't have to all live in the same Parquet file. you can have a Parquet file for all your historical data, plus one for the current week which is updated often cheaply, and then when the week is over you merge that into your big parquet file.

You're making it seem like there's hard limits to what can be done but while there definitely is, you can do incredible stuff.

Being on the HN's homepage isn't as simple as "most upvoted articles in last X hours". They take into account upvote velocity: 40 upvotes in 40 minutes isn't the same as 40 upvotes in 4 yours.

No mention of WebTransport makes me very doubtful of this article. It's 2026, all browsers support it and it's bidirectional and lower-latency than WebSocket.

Well, you tell your Roomba "clean the floor" but you don't ask an AI "make a game". You give it very specific instructions.

People complain when a product use a familiar name that might collide and there's also people complaining when they invent new words altogether.

Naming things is hard.

What makes you think I don't care at all about facts or the truth ?

They get all our invaluable data which they'll use to train the next model, to get more data, to train the model after.

Why should I need to do any testing ? I'm just going to keep on not believing you until you back up your claims.

And your iPad example doesn't make sense at all, you compared 2 Macbooks and now you're talking about another OS altogether.

> idk what to say.

You could just tell us one example of something that's faster on your 2019 Macbook Pro 15 versus a M5 Macbook pro and how to reproduce it.

Which tasks ? I don't believe your claims.

You can disable it in the accessibility preferences.

Back in the days we used to call this R&D.

Warning: this link automatically downloads a .dmg file without asking.

https://screenpipe.com /how-to-install

> But that is what people play

Lots of people don't play them. It's not like we don't have a choice.

It'll most likely land on Bedrock when the K3 weights are released.

The thing is that if your config language isn't Turing complete, the community will create N softwares to generate dynamic configs.

If something is hard to measure, is it automatically bad in your eyes ?

Anthropic makes more money when people use 5% of what their subscription offers them. This allows them to sell more subscriptions without paying for more capacity.

We're taming a beast.

It's a mix of french and english... so it's german.

I've had Fable read and write code. Never saw any downgrades.

Absolutely. This is DOA honestly and not really better than what we had before.

It's coming in 7.1

I've done many custom low token output CLIs like this for my day job and it's something I expect to see much more of.

They were faking artificial intelligence by using real individuals.

It now drives the "actors in this scene" feature in Amazon Prime.

Article name: "everything the doc doesn't mention"

You: "they missed this feature that's in the docs !"

You're right that it's an important part of CC's config. But it doesn't fit the article's raison d'être.

AI versus AI, the final faceoff. Who's gonna win? Probably not us.