verytrivial

Honestly read this as satirical after my brain inserted "intelligence" after reef. It certainly feels like that.

I must say this is the sort of war I can get behind, where one side seems to be offering us ever fluffier socks and back rubs and the other shouts into the wind that only their flavor of wholesale intellectual theft is moral.

Wow. I've tried a bunch of agent harnesses and Codex Desktop was on my list to try. It has definitely fallen off!

SpaceX have great inertial measurement units and cold gas thrusters. We get it. Musk remains a fascist and can keep his cars.

Every uncoated pill that has a curved surface can be split by placing it on a hard surface and pressing down on opposite sides with your thumbs. The split is more accurate if it is pre-scored -- place the scoring/groove face up during the split. But even without the groove, it is rarely worse than a 55:45 split in my experience. And no, I have no idea why this story is on HN.

It's always a gamble to attempt humour here. Try the veal.

People have been hallucinating meaning in whale song for decades. Why not automate it?

Strong smell of Ai generation on the cover there.

Best comparison that has occurred to me is the cost a loaf of bread's ingredients might be slightly cheaper than a baked loaf, depending on how you source it. At home you get total control and know what's going in to it. Yet bake at home is still a niche, perhaps a hobby. So I say as someone who's spent hundreds of hours tinkering with local inference, go for it for anyone reading. But most people just want ... some slices of bread, you know?

I find it interesting that with the given metric comparison, for coding at min 50 strength, every frontier model brand is from a distinct vendor: Ling, Qwen, Gemini, Muse, Grok, GPT, and Claude in increasing value.

In a corp environment, it is the business's problem when a user's passkey devices fails. In a personal environment, it is the user's problem. Specialized knowledge and preventive processes in the corp environment aren't present in the personal context, and saying "pretty please" to Google or Apple when something goes wrong is usually a waste of time. I 100% agree with TFA -- "damage expectation" is probably higher with passkeys.

There's a chap called Bijian Bowen who does very quick agentic coding challenges for new models (very soon after release!) mainly for toy games or websites. He just did one for this model and included a comparison with the base model Qwen 3.8 which shows the "near-lossless" claim should be taken with a grain of salt. It is an interesting model if you are GPU starved and want local, but you might have trouble finding things it is good at.