What's impeding a lab from releasing its own new model, pricing it really low for the beautiful Pareto plot, accompanied by phrases VC love like "establishing a new frontier in cost", to just then raise prices back up?
They don't even have to "raise the prices back up". They just need to ensure that the number of output tokens goes up needlessly. Just make the model more talkative (via RL?) and cha-ching!
Interesting. Perhaps I can see this being quickly adopted in LLMs-as-a-judge, where you normally need (a) a structured answer, say, with lots of different fields (metrics) and (b) you want the judge to be fast, not being a bottleneck.
The CTO himself reported 600 commits and 300 PRs each day. And they can do so because they have a massive CI [1]:
> Every day CI runs about 20..80 million tests in 600 commits and 300 pull requests
> Last year, ClickHouse spent 360 years of machine time for CI
I am no user of CH so I can't talk about their product. But we are talking about a company with 686 employees as per their LinkedIn, where ClickHouse is clearly the core of their business. Considering all of this, is 50+ commits a day that much?
I think you're right. We're at the point where Apple is sort of out of steam in terms of hardware and software and their products are mature and the company has become more risk averse. They need gimmicks to get people to upgrade and I'm sure we will see more attempts like this in the future. These sorts of devices have limited use cases, like people who are on the road a lot for work or leisure, and of course this appeals to tech enthusiasts or the vanity tech crowd.
It'll be a nice device for some, but I think we're at the stage where device upgrades are no longer going to be a yearly or even biennial thing with all things considered.
Everyone in my family wants one... assuming it works well. We spend a lot of time on our phones, and a lot watching video. This seems like the ideal device for that.
I have no doubts this device will be in demand and sell a ton of units
But what are we even doing? $2000 for a device to watch videos on, to replace the $1200 device you watched videos on last year.
I guess I’m lamenting the fact that none of this amazing technology is actually doing much for us any more. It wasn’t that long ago that every year we would get new devices that actually changed how we did our day-to-day
I think there's plenty of interest, Samsung, Pixel and Motorola are just some of the companies with multiple generations of foldables. Besides enthusiasts, the type of person that might want an iPhone duo is someone that doesn't really use a laptop. The utility of a tablet for media consumption and light productivity with a small footprint is real thing.
The problems besides price (which are still insane) until now where:
- Materials not catching up, things like the older creases, dust ingress, etc.
- Software in Android being hit or miss, specially on with third parties. Apple has the muscle to make apps compatible with the form factor.
- The "standard" foldable aspect ratio sucking ass, the only major exceptions being the Surface Duo, the original Pixel Fold and the recent Galaxy Fold. Passport format is simply better.
>Imagine if Apple dedicated this huge engineering effort onto existing products.
For what? Slab phones are practically solved, the only things that might make a normal iPhone better besides incrementally better camera, battery and processors are not Apple approved. No expandable storage or cheap battery replacements for you. The only place for real advancemnt will be foldables for a while.
Reminds me, kinda, to when Astra was launched and OpenAI announced an improvement to the bounded prime gap. Which BTW, Prof. Julia Stadlmann had published an independent result only a few days earlier
Stadlmann improved it from 246 to 240, OpenAI later claimed 186 I think?
Maybe someone can help clarify? I am no expert at all, but I can't help but see similarities.
Is it surprising that different groups are working on the same problems? With each new model generation, the LLMs get good enough to solve a new small fraction of open problems. Of course the problems that get solved are going to be the same subset.
2026 has been the year where spec-dec has matured, it has been adopted by all big OSS engines and i'm sure it's present in quite a lot of inference providers as the default
I feel like P/D dissaggregation will be the next big one for providers, as prefill tends to be compute bound while decode mem bound which I guess each will have a different type of node
reply