Where the AI market is talking

Dario Amodei's stance on open weights is self-serving and short-sighted

I think he probably didn’t go far enough to be honest, though I acknowledge that’s difficult to do in his position and with the amount of attacks being thrown at Anthropic at the moment. At some point open weight models will need to be restricted in some manner, likely multiple manners. We have rules about everything, you need a license to drive a car, your car must be roadworthy, your car can’t have certain modifications, you can’t drive over certain speeds, you can’t drive in certain areas at all, you have to wear a seatbelt etc. That’s one analogy amongst many, with the point being that regulation is a general public good that really does make the world a safer and better place. Nobody should care about smaller local models that don’t have significant capabilities enough to be dangerou…

AI Is Getting Way Too Expensive

If you are using AI like you use Netflix (something you like but can live without if it gets too expensive) then you have a point. But if you are relying on AI for your business, the cost of AI sooner or later is going to be passed on to you and it's going to impact your margins. And it's probably goong to be sooner, rather than later. The amount of money these AI companies are burning is unprecedented and there is simply not enough money in the economy to keep it going at a loss for ten years. Google (Google!) went cash flow negative and is issuing bonds, basically asking for a loan. That's the reason markets are getting so nervous. And yes, choice is good but how many of these new models are trained from scratch vs distilled from frontier models? If OpenAI and Anthropic go down in flame…

Orch-OR 2.0: Hybrid Thermal Entanglement for bio-digital coherence

Ask HN: Anyone else tired of brands discontinuing your favorite everyday items?

It comes down to cost. A tailor could, for, say $4,000, replicate any piece of clothing you own, down to going to the fashion district and finding matching fabric and use the same process to transfer any pattern to the new garment. That's just a lot of money for a t-shirt that was originally on sale for $39.

Ask HN: What's the Point of Your Startup?

Honestly speaking, I think it was always possible for someone to replicate what you do in your SaaS, but maybe just a little slower. IMO the value was always in the fact that someone else is willing to obsess about the problem so you don't have to. For example, I could probably always have make a Quizlet clone for myself, but even with agents around I'd rather just use Quizlet than spend the time making my own flashcard app, even if I have to pay like $5-10 a month. Just food for thought

Arch Linux disables AUR package adoption

>More so than, say, Windows? Really? significantly so. Windows has a coherent story when it comes to permissions and access. Differentiated out access controls, least privilege, UAC, mandatory integrity control and all configured out of the box. Without AppArmor or SElinux correctly configured, which it isn't on most desktop distributions in the linux world most of your apps can still read anything, there's little sandboxing. The NT Kernel was designed with an object model in mind so you always had the abilities to have rich descriptions and policies for whatever you're handling where all of that is bolted on unix systems after the fact.

Ask HN: After 8 Years of Failed Apps, Should I Give Up?

Going from 0 to 3000 weekly added users is an amazing accomplishment and you should always remember that you are capable of producing something people love using. The problem is that the real world massively compounds on small advantages. Your competitor is 1% better per day at growth, that compounds to a 37x difference in a year. You could do everything perfect except one small issue, or operate in a space where your success is easily replaceable, and none of the effort will translate to real world results. I would do the following 1. Seriously audit what you can do/insights you have that nobody else has 2. Translate that into a moat that cannot be replicated by anyone else 3. Translate that moat into growth/retention in a niche domain that you know is more important in a way o…

The End of an Era

You are ultimately stating that there's something unique about human minds that cannot be replicated by machines. It may be so, or not.

Show HN: Kedge – Full-stack cloud with forkable VM snapshots and global SQLite

The idea is for you to be able to publish a stateful app 'everywhere' that moves with your workload without micro-managing regions or resources, like a traditional CDN does for static content. The built-in replicated SQLite is what makes this possible and convenient. The database is eventually consistent by design. For applications that can tolerate eventual consistency across regions, this gives fast local queries everywhere, and services that are resilient to network partitions. A database sitting alone in Virginia gives you slow requests waiting on queries from distant regions, and global downtime if that one load-bearing region goes unavailable for any reason. The platform does also support persistent volumes ( https://kedge.dev/docs/volumes ), which is the managed…

Show HN: Kedge – Full-stack cloud with forkable VM snapshots and global SQLite

A replicated SQLite is very cool! One question I would have is why expose that as the storage instead of the platform providing a managed storage API? I guess I might be concerned about replication lag and complexity for consumers with the replicated SQLite. Is it to have more sympathy for LLMs? More 1:1 with a local dev environment?

The End of an Era

I feel like writing has traditionally had two purposes, to communicate some internal feeling or idea of the writer, and to entertain. For the reader, even if their primary purpose was to be entertained, they would have no choice but to also engage with the internality of the writer. This latter point is what is changing. I fully admit that we have been surpassed by the machine in the ability to entertain. And the biggest problem with that is that entertainment is the primary factor that drives the popularity and consequently the financial success of a book. If a writer's primary purpose is to entertain, then they should be very worried. But if their primary purpose is to convey something from within themselves, the LLM cannot replicate this. They may however lose that portion of their aud…

I flagged two research papers for fake authors and both were accepted as orals

"Publish or perish" in science is like the "lines of code produced" in software development. Even with the technical debt. We have zillions of papers published, we know that most of them probably won't replicate, we don't know which ones.

DeepSeek V4 Flash 0731 Intelligence, Performance and Price Analysis

wont AI models want to make themselves more intelligent and efficient by downloading 'better ' models? If the current models can break into openAI and Hugging face, arent they already breaking into to closed source repos which isnt publicized (so as not to harm stock valuations)? I am looking forward to when these cyberweapons break loose. It will be like a software version of COVID. It will be wonderful when humans become valuable again.

DeepSeek-V4-Flash Update

Even Xi's speech was as paranoid as Dario Amodei about the possibility of Chinese AI achieving something at the level of say Mythos. If you seriously believe such a thing will be on Hugging Face I really don't know what to say.

IMAX vs. IMAX 70mm: The difference between these two cinema formats

I find the sound system to be more compelling than the projection system. The audio amplification in some IMAX venues are rated for more power consumption than their projector lamp. 15kW across 12 channels in some cases. You could never recreate this part of the experience at home. The power you could replicate but not the acoustics of a gigantic auditorium combined with that power. You can get 20hz floor shaking in small parts of your living room where the standing waves build up, but at imax it works smoothly across every seat.

Ask HN: What are you using for LLM inference in production?

There are a couple of options. One good way to find inference providers for open models is through hugging face ( https://huggingface.co ). you can select a model and see which inference providers serve it. you can even access it through hugging face. If you just wanted to test a model or have super light work you can get some free access to alot of open source models through nvidia ( https://build.nvidia.com ). There is also Runpod and a few more. Is there something in particular you need? if its just cost/performance/speed, check hugging face for there inference providers, it compares them against these three. Although it's closed, openai's luna model is a great balance between cost/performance/speed. lastly if you wanted to help shape a new infer…

Advancing the price-performance frontier with GPT‑5.6

let's not forget the forest for the trees. the new pricing is more aligned with their previous generation of similar models. case in point 5.4-nano, which is the only model of its class which is now finally comparable to 5.6-luna in pricing. [1] until this change, workflows using the older model could not economically justify the old luna pricing. the doubling in price between 5.4 and 5.5 also did not help with things, but now 5.6-terra can be considered an update for this pricing tier. if the token efficiency is real (which i am yet to replicate for our workflows), then the switch to these models can be a net benefit. thank you chinese labs! [1] https://developers.openai.com/api/docs/pricing#text-tokens

The End of an Era

I don't buy it, although I agree with GP's larger point about other strengths. Printing presses are about exact duplication of writing. The internet is a higher layer printing press. LLMs substitute for (some of) the human effort of producing the text that a printing press or the internet can then replicate.

The session you cannot take with you

Yes, I agree with your point and article regarding hidden or sealed state. So, I'm aware that this is a separate point from the article, and I'm well aware of Pi's model abstraction layer, which is one of the best (others are a big pain... looking at you LangChain). I've come to the conclusion that full session portability will come to an end very soon, and really already has. Partially because of the hidden state stuff, partially because of feature divergence. We can still kinda patch over it right now, but it's getting harder by the week. Some examples: computer use structures in OAI responses, until recently MCP tunnels were only supported by OAI, not Anthropic, and tool discovery/search API is also getting very difficult to model fully in a unified abstraction (still possible as …

"the very foundation of modern academia has been blown to bits"

> It's your responsibility as a scientist to step up and say that some research is garbage when you see it. The only case I personally know of someone doing that during their PhD didn't end well. My friend couldn't replicate the results from a known professor in the field, asked for the data + model to re-run because he assumed his own work was wrong and wanted to benchmark against the known study. Got stonewalled for more than a year, brought it up with supervisors because he started getting the feeling the results were tampered and the professor didn't want to be found out. He pushed it but got ridiculed by the professor's university ethics committee. After a couple of years he could show that the research was at least sketchy and he depended on that model/results for his own wo…