I also struggle to reconcile these things, I guess the only way would be to try and see if it works for me, without caring too much if it works for other people on the internet. I would just like to add that, DwarfStar's author (Salvatore Sanfilippo) is a strong supporter of the idea that you shouldn't read the code, and he says he never read DwarfStar's code. And still, it seems that this project is much more than PoC and actually both usable and useful for people doing local inference (I didn't try it myself, but I saw a lot of positive comments about it). Could be that the crucial point is in how we use those models: instead of giving it a general goal (e.g. build me an inference engine) Sanfilippo, being an experienced programmer, kept pointing the models in the right direction. He also read the papers related to the models he was programming support for in DwarfStar, so that, when he worked on optimizations, he knew what should be done instead of prompting a general "please optimize this". So, I would say that, if you let the agents work on a "feature by feature" basis instead of trying to on3-shot things, you get much better results. Could also be that, by attempting to one-shot large projects, the model starts coding badly due to context window exhaustion.
Sorry for the not so well written comment, I was just throwing in some ideas.
Yes, you can get better pricing if you do it yourself. But the true advantage of OpenRouter is that, in a space where there's a new model being released every week, you can easily switch to whatever model is best at any given time without having to set up accounts with multiple providers. Or you can just experiment with the latest release. Their product is the convenience. Of course if you decide to only use a specific provider or two, then you don't need OpenRouter.
But it's not really that difficult to make a clone of OpenRouter's service. What they do isn't really that original.
Their only value comes from the fact that the currently have lots of traffic. And I dkn't think that their cumstomers are really bound to theur servuce. They could switch to a competitor without too much hassle.
Yes, but as l9ng as there's no clone, they're good. On the potential clone side, I guess many will be put off by the fact that "there's already OpenRouter". Also, a clone would need to find a way to make existing OpenRouter customers to switch, which isn't easy.
One of my sci-fi future ideas, every time I hear about someone thinking of a new way to make warp drives, is that if we ever sort it out someone will make TARDISes for the filthy rich - a static warp bubble that moves space into a ground floor closet or room from space in an upstairs or in the attic.
Like a milder version of the gimmick in Ultraviolet.
IMO there's a big difference between using AI to write code and using it to review your code. In the latter, you know the code well (you are the one who wrote it), and so it is much easier to understand if AI suggestions are good or hallucinations. In the former case, realizing when the AI is making mistakes is harder, because you don't have the full context anymore.
You should really look at studies on how happy employees are much more productive than unhappy ones. And the trend continues way above the bare minimum of not being miserable.
not sure why you've equated the team's happiness is less important than the software to make the team unhappy. that's an incredibly binary perspective.
there's usually more than two priorities. and i agree with you that a team's wellbeing is important. it should be up there in priorities.
i'm just pointing out that a team that accepts being <some imaginary figure that's not 100% like 70% or 80%> happy, focussing on the software as more important than their own happiness might do better than a team who are trying to focus on their happiness at the same level of priority as the quality of the software.
I agree it's not a binary thing, it's just that, by how the parent comment is worded, looks like for them anything above the line of misery is OK. But maybe my interpretation is wrong.
> We plan to raise the overall pricing for DeepSeek API services in the near future, with a significant increase expected. Please plan your usage accordingly. The specific pricing plan will be subject to official notice.
It would be nice to have this notice on the public documentation as well.
The basic services that keep us alive (e.g. farmers producing our food) tend to also be "easy" (it is physically demanding, but doesn't require years of study to start and you quickly learn by doing). This also means that the vast majority of people would be able to do this kind of jobs. Our economy pays more for scarcity, i.e. jobs that are in demand and at the same time require more qualification, so that the number of people needed in that position tends to be less than the current amount. Both views have sense, they just start from two different points.
From having worked with farmers, the actual logistics of running a farm are very complex. Farming's no longer an easy job because the demand for productivity means you have to harness a whole stable of technologies and understand to some degree advanced finances, floral biology, geotechnics, and meteorology. And yet farmers still get paid little relative to the breadth of knowledge and skills required for running a farm. It's just that farming is still considered a blue-collar job because it was the domain of the poorest for millenia, where as software engineering in any form is relatively new and thus specialized and so considered white-collar.
You're right, running now a farm that is profitable and can stay on the market is very different from doing so 200 years ago. But knowledge is just one aspect. The other aspect I mentioned is having less people in the field than what the market asks for. In the specific case of farmers, I think that modern farms are so much more efficient that we need way less than before, even if the world's population grew. So, while there's a gap between the knowledge required and the earnings, I think this can be explained by farms having a much greater output now, and so the market doesn't ask for more farms than what we have. This could be false at a global scale, but I think it is true at a country-level scale in most of the developed countries. Hence why farmers aren't paid much in those countries.
Farmers aren't really paid, they sell goods at market prices. Modern farms are more similar to operating a manufacturing business than they are to a salaried position or profession. They don't earn high incomes compared to white collar professions but they do have high revenues and they pay themselves a reasonable middle class income on average with some added benefits from owning large assets and land. The important part about comparing farmers to programmers is that they are both now operating in highly competitive markets with high automation where a single individual can leverage their knowledge and ability using that automation to produce many times more what a person could using only their own labor.
> Farmers aren't really paid, they sell goods at market prices.
"Programmers aren't really paid, they sell services at market prices." — I guess nobody gets paid?
As a farmer myself, when I give someone something I produced and they give me back something in return, I consider myself to be paid. If I did not receive much value in return for what I gave, I'd reasonably consider myself to have not been paid well. I have never met a farmer that would consider that usage unusual. Payment refers to a transaction within the realm of a human activity. The good or service that is being traded between humans is immaterial. While you are free to define words as you see fit, your assertion doesn't really reflect how this is actually used in practice, generally. The rest of the earlier comment makes it clear that the commenter understands how a farm business works.
Did someone seriously just down-mod a farmer in a sub-discussion about running a farm?
PS To the "clear-thinking" guy at the top of this thread...this is why...you only think your thinking is clear. If its a topic you know nothing about, devs (not necessarily you) tend to assume there isn't anything to know and that's almost never true.
No, nobody would seriously press a button that does nothing but increment/decrement an integer. Imagine the thought process that would have to go into thinking that was something worthy of one's time and attention. Not even those who struggle with clear thinking would be able to come up with a story so outlandish and contrived as to start seeing it as an activity to want to get involved in.
It is easy to accidentally press the button and not notice, though.
Most programmers are of the capital holding class too. At very least they will own a computer, which is all the capital you need for 99% of what programmers do. There are exceptions, but there are also exceptions in farming.
Land isn't traditionally considered capital. The buckets are normally land, labor, and capital. The closest thing we broadly see to someone whose function is holding capital is someone who owns stock in the stock market. It is likely that most programmers own stock. If I were a betting man, I expect that programmers have more impact on capital than farmers do. But we still have to split hairs over what "main" means.
One interpretation, which is what I suspect most would go to, is where one's income is primarily derived from owning stock. So, in other words, old, retired people. But I don't think anyone here was thinking about the retired. The discussion was about people actively producing food and software.
Another possible interpretation is about impact. If those stocks didn't exist, a lot of the capital wouldn't have been able to be created, and thus there wouldn't have been jobs to go with it; meaning that owning the capital makes it one's main function as without it one wouldn't have a job to go to. But I can already hear you furiously typing how that isn't a satisfactory take. I'd agree.
Maybe it's about attention? Someone who actively watches over to their capital would be seen as it being their main function, while someone who leaves their capital ownership up to a financial advisor so they don't have to think about it wouldn't. But then we're back to the computer. Most programmers watch over that capital, if nothing else. That doesn't really work either.
The only remaining interpretation I can come up, given the connotations often associated with landlord, is a group of people not like by another group of people. But that doesn't have anything to do with capital at all, so... Consider me stumped on what this has to do with the discussion (if we are to ignore the explicit sarcasm involved).
You are not technically wrong that our modern technology has made it possible for land to be capital, like when you dredged the sea floor to build an island, but is a tiny, insignificant portion of all the land. If you are under the impression someone needs a lesson in economics, what are the chances that they are going to think you are talking about some man-made island in Dubai rather than the natural ground that makes up the vast, vast, vast, vast majority of earth? For that explanation to land the audience would, in most cases, need an even deeper understanding of economics than the level you were trying to present.
Isn't Elon Musk's main function in society court jester — to make a fool of himself for our entertainment? That is what he does in my timeline. Perhaps we exist in different universes.
"The basic services that keep us alive (e.g. farmers producing our food) tend to also be "easy""
The reason you think that is that you know nothing about agriculture. But it isn't true. Right now, farmers in the US grow about 5x as much food per acre as farmers in other countries. If you were in charge of farming in the US, that surplus would likely disappear in a single season. You would also likely damage the soil which would make cleaning up after you take 4-6 years.
Everyone thinks things they know nothing about are easy. But they rarely are.
PS This is why I never tell people I'm a programmer. I don't want to be associated with people who think like you.
You should read my other comment where I explain better what I meant. I also don't like how some programmers like you tend to think they're superior to others ;)
Strange that in the prior art they didn't list DwarfStar, as it is able to run the same model (probably quantized differently though) in less memory. Maybe the author isn't aware of it?
We can only wait to see if this is true. Another possibility could be that it was an answer to USA's government considering a ban on chinese models. In this way, he fueled the discussion around the importance of open weight models.
Xi actually said that models of the strength to pose security issues will be permanent state secrets -- in the same speech that the first western takes translated as actually using the words 'open weights' which of course nowhere appeared. Now: when will "models of the strength to pose security issues" appear? Maybe my inference 'months' is wrong, but they seem to be advancing rapidly.
The administration blather about banning open weights is characteristically confused. Xi has already stated (what is obvious) that he will ban security-endangering weights and keep them a state secret.
Sorry for the not so well written comment, I was just throwing in some ideas.
reply