Exo-Labs is an open source project that allows this too, pipeline parallelism I mean not the latter, and it's device agnostic meaning you can daisy-chain anything you have that has memory and the implementation will intelligently shard model layers across them, though its slow but scales linearly with concurrent requests.
Last year o3 high did 88% on ARC-AGI 1 at more than $4,000/task. This model at its X high configuration scores 90.5% at just $11,64 per task.
General intelligence has ridiculously gotten less expensive. I don't know if it's because of compute and energy abundance,or attention mechanisms improving in efficiency or both but we have to acknowledge the bigger picture and relative prices.
Sure, but the reason I'm confused by the pricing is that the pricing doesn't exist in a vacuum.
Pro barely performs better than Thinking in OpenAI's published numbers, but comes at ~10x the price with an explicit disclaimer that it's slow on the order of minutes.
If the published performance numbers are accurate, it seems like it'd be incredibly difficult to justify the premium.
At least on the surface level, it looks like it exists mostly to juice benchmark claims.
It could be using the same early trick of Grok (at least in the earlier versions) that they boot 10 agents who work on the problem in parallel and then get a consensus on the answer. This would explain the price and the latency.
Essentially a newbie trick that works really well but not efficient, but still looking like it's amazing breakthrough.
(if someone knows the actual implementation I'm curious)
If Arabic had to cater to afro-asiatic dialects phonemes then the script would have been even more messier. I'm a speaker of one, and my dialect is heavily influenced by the indigenous Tamazight language. and I think this is why many of the Amazigh community were and some still disappointed with the neo-Tifinagh script. While it carries symbolic weight, it doesn’t offer practical readability, phonemic clarity and tech accessibility of a modern script that Tamazight deserves. Latin script, ironically, fits Tamazight much more naturally.
4.8b words on English Wikipedia. Knowledge cutoff of 6 months. A valid use case is to search across Wikipedia and ground your
answers.
Trivially proves that RAG is still needed.
This is only for the small model. The medium model is still at 1M (like Gemini 2.5)
Even if we could get the mid models to 10M, that's still a medium-sized repo at best. Repos size growth will also accelerate as LLMs generate more code. There's no way to catch up.
I don't know what differentiates them since the content is apparently similar but I suspect a lot of websites like this will diversify their distribution channels and pivot accordingly.
>China is not the threat to Europe that Russia is. A wedge between China and Russia would weaken Russia and help the EU
It gets interesting when you realize that Russia is also a rival to China in Northeast Asia. A balkanized Russia, like the one the EU could have manifested had it took Russia warnings seriously and brought about decisive action after troops were invading Crimea. But no they lived in their "End of History" fantasy and that virtuous liberties will magically be spread if we just trade goods and ideas between spheres of influence.
Of course this reality will be bad for our allies in Asia (ie. Japan, SK, Taiwan). But maybe this time it'll wake up some in America from becoming isolationist again.
I see SK and Japan also as necessary allies for the EU. But if China decides to take Taiwan, I don't think there's anything the EU could possibly hope to do about it; Taiwan remains independent because the US guarantees their independence. If Trump were to withdraw that guarantee too, I don't think there's anything that can save Taiwan.
It sucks, but the EU has more urgent problems closer to home. All I can hope for is that Trump hates China enough that he'll continue to guarantee Taiwan's freedom. But I'm sure at some point he's going to ask them for some more material "thanks" too.
But yeah, the EU's relationship with China should not be the same as that with other allies. But I think there's room for some cooperation, and the EU might not object too loudly if China were to take outer Manchuria back, for example.
Honestly I don't think I can see China taking Taiwan militarily. They witnessed how the world isolated Russia economically after the aggression on Ukraine and especially since their economy relies on exports. They don't have much to win other than some geopolitical credit at the expense of their manufacturing and technology sector.
China is conducting a policy where they'll cripple Taiwan's will to seek independence from just sheer soft and economic power. They offer fantastic perks to Taiwanese from the oppurtinity to work visa free, access to credit/mortgage with no social credit screening and ability to invest with no usual red tape.
The Taiwanese are being told China is an aggressor but nowadays they see the opposite. Also if China invades it'll destroy every goodwill they had built to win over Taiwanese hearts and won't get control over TSMC supply chain market since the latter promised to torpefy their fabs before China gets its hand on them.
>In 20, 15, or 7 years from now when terrorists are sending drones into medium-sized cities in Alabama to kill indiscriminately, it would have been better for the USA to have been on Ukraine's side.
Spot on. This is what Zelenskyy implied when he said "now you have an ocean but one day you'll know how it feels". But the dumb kakistocrat commander-in-chief took it personally.
By the way remember the New Jersey drone sightings that spooked the East Coast for a week? That was likely the government secretly testing defense deployement against a hypothetical drone swarms attack.
I have assumed the same about the gatwick airport shutdown caused by drones in the UK in 2018. Newspapers saying "No culprit ever found" says "results of military exercise are classified" to me...
IMO if there is any model to follow to solve the housing crisis it should be Japan/Singapore. Vienna is too unique of a case and its housing market still suffers from negative externalities.
Singapore is also hard to achieve but the HDB mandate made it that the government own monopoly on land so they're incentivized to provide maximum land utility to its population.
Japan has the best land use zoning in a way it makes it so easy to build and develop. They also have a very practical view on housing and do not see it as an appreciating asset critical to accumulating wealth.
> Japan has the best land use zoning in a way it's makes it so easy to build and develop. They also have a very practical view on housing and do not see it as an appreciating asset critical to accumulating wealth.
I want to note that this is in big part due to earthquakes and nature in general. In Spain 100-year old houses (like in the article) are renovated and relatively easy to maintain. However in Japan, between the earthquakes, typhoons, and the weather in general, not only you need to be actively maintaining it (typhoons), but they become worse over time (earthquake damage accumulates).
Yeah that would be a factor but there's more to that. For example the Japanese are so crafty with land use they even allow an odd type of development called Zakkyo. Which are multi-storey buildings wherein each floor is owned and managed independently. The foot traffic isn't restricted to the ground floor to enjoy these "third places".
IMO geographical constraints should not be the all-in justification for a certain land-use/housing policy. I hear a lot chalk up the fact Barcelona stopped building since the 1980s because they're surrounded by mountains and the sea. Or recently with the LA fires people look at these wrong answers. There's no reason for things being like that other than NIMBYism and lack of long-term political will from local councils.
Spain's multi-story buildings are often independently owned anyway: the "law of horizontal property", dealing with condominiums, is there to mostly deal with shared disputes among owners. Buildings where the ground floor is stores, the next three have small professional offices, all owning the space, and the rest is a collection of apartments occupied by their owners is very common where there's sufficient demand for said offices.
Barcelona's problems come from a very complicated regulatory regime that hasn't just lowered construction, but ultimately led to fewer people living in the same space. Regulations that favor the renter enough to make putting housing up for rent pretty risky in many ways. Minimal property taxes that let having an apartment unused be pretty attractive. Big tax advantages for owning an expensive apartment vs living in a smaller one and putting your savings into stocks. Prices that don't come down. A location so interesting, and yet at prices that are globally affordable, so it is more profitable to have an apartment dedicated to low-occupancy rate foreign tourism than to have it occupied all year long by a long term tenant. Cheap enough for an American in tech to buy an apartment downtown, and use it 3 weeks a year.
The market-centric changes that wouldn't just allow more building, but make the existing buildings be used more efficiently are just not in any local regulator's radar, because lowering prices might be nice for renters, but it'd be awful for owners, and there's a lot of those, and they vote.
Definitely, but I wanted to note a big thing that I often do not see mentioned in these kind of discussions. e.g. I'm looking at apartments in Tokyo, and for Japanese people one of the biggest decision factor is the age due to earthquake fears. Heck, I'd say it's the main factor not to buy, a big % of the people just won't buy older than X years. This is the most "unthinkable" thing in my Spaniard brain, since older houses are not inherently worse (they have both bad things and good things).
There's many other differences (and surprisingly, similarities) between Spain and Japan regarding buildings, e.g. the mixed areas are both very similar in both countries. But another surprising thing in (specifically) Tokyo is how strongly public transportation shapes Tokyo, go out in any train station and you have tall buildings, but walk 10-15 mins away and you have houses. In Spain, you have a very homogeneous height all across the city (fun fact: this was traditionally shaped by the cathedrals/churches).
