While I understand both of your perspectives, please think about how frustrating it can be to have poured years into building a product genuinely solving a problem and building a relationship with clients only for a dozens AI copy cats to come in. Even if your product is clearly better, the SEO slop alone will raise your marketing cost, price cutting will affect you as not all clients are able to tell quality before purchase or will know of your project, etc.
And in the end, everyone is worse of. We get less for the products we build. Customers in many cases get a worse solution. And the AI copy cats multiply until none of them can make a living from it.
If your product is really good, then people will choose it over AI copy cats. If that’s not the case then it’s time to take a really hard look at your product and see if that’s really as good as you think
That's a naive view of things. People don't have perfect information about the world around them. They can only go on what limited information filters to them. AI copycats will by their very nature be better at marketing and pushing information into people's faces. It's bad information, but how are customers supposed to know that?
Exactly. People on a forum dedicated to a VC should know that disruption happens everywhere, AI just makes it easier but it's been happening for a long time, that is the nature of startups in general.
I don't know, it really depends on the product. If we were all still paying for proprietary UNIX in the current year, then it's likely that these startups would never exist in the first place. Sometimes a SAAS has to admit that it's not providing real value before actual innovation becomes the status quo. I don't weep many tears for dying businesses because nobody lives forever.
That's roughly 11.66$ per hour per GPU, which is above the price that Google Cloud is reselling them at if you commit for 36 months: https://getdeploying.com/gpus/nvidia-b200
That's for standalone servers. xAI has something that nobody else has. The single largest interconnected cluster[1]. For inference it doesn't really matter, but for training that networking is crucial.
[1] Probably, there could be undisclosed clusters owned by other companies.
There are definitely more shades of grey. On my iPhone I can select a close contact to be able to overturn my protection but this contact needs to have security features turned on, too. So Apple staff cannot do it, only a non publicly known person that has 2FA and encryption themselves. Add time delays, notifications, identity checks and more to it and you can make this process reasonably secure while still ensuring recovery.
It's discouraging to see Google price Gemini 3.5 Flash at 3x the cost of Gemini 3 Flash. I would think that most people that deployed this model in production would have used it for low latency tasks, classification/categorization, customer support or basic RAG-/RAG-style chatbots. Performance on coding benchmarks is nice and all, but where is the "intelligence too cheap to measure"? This new cost point is quite prohibitive and will eat up a lot of margins if developers adopt it.
Expect all models to increase in price 3x with new releases. They're easing us into the margins they're targeting.
Flash 3 wasnt appropriately priced, it was priced to get you used to a certain level of spending, then they'll crank it up and get you used to the next level of spending.
I am aware that it was likely subsidized or at least did not have appropriate margins. But over time, that same capability should become profitable if parameter efficiency and chips improve. For many customer facing use cases outside of coding assistants, optimizing for speed, basic logic/maths and conversational texts matters much more than being able to use 40 tools simultaneously. I would have hoped that Google would recognize this and keep a dual line up, where Pro and Flash are clearly intended for different market segments. But it seems, it's all in on coding assistants and screw the other use cases..
Now, we might need to change to DeepSeek 4 Flash if Google deprecates 3 Flash.
When using Hetzner, DigitalOcean or any other VPS service together with Cloudflare, I can handle millions of page views for 5-50$ a month at pricing that has stayed nominally the same for a long time and due to inflation and performance gains of the underlying chips has basically become cheaper.
I guess you didn't get the memo from last month: Loss leader pricing is over, you're now paying a less subsidized price, and will continue to until it's profitable
As explained in another comment, I think this is more about Google orienting Flash towards more complex use cases. If we got minor improvements vs 3 Flash with 1.5x the price so they can optimize their margin (which on such small models for conversational tasks is a completely different stories than the 3-25x subsidies that these agentic coding plans offer) I would have been happy. Or even no change at all. But knowing Google, I now must fear that they will deprecate 3 Flash without offering any realistic option for that user facing chatbot segment that does not require multi-tool use across 500k context.
I mean, the 100$ plan is less than the hourly rate of any consultant / senior dev in developed countries. So if it can save even one hour a month, it's cost efficient for the customer (at the current, subsidized rates, of course).
I came here to say something similar. As someone who works in a field that applies machine learning but is not purely focused on it, I interact with people who think that arXiv is the only relevant platform and that they don't need to submit their work to any journal, as well as people who still think that preprints don't count at all and that data isn't published until it's printed in an academic journal. It can feel like a clash of worlds.
I think both sides could learn from the other. In the case of ML, I understand the desire to move fast and that average time to publication of 250-300 days in some of the top-tier journals can feel like an unnecessary burden. But having been on both sides of peer review, there is value to the system and it has made for better work.
Not doing any of it follows the same spirit as not benchmarking your approach against more than maybe one alternative and that already as an after-thought. Or benchmaxxing but not exploring the actual real-world consequences, time and cost trade offs, etc.
Now, is academic publishing perfect? Of course not, very very far from it. It desperately needs to be reformed to keep it economically accessible, time efficient for both authors, editors and peer reviewers and to prevent the "hot topic of the day" from dominating journals and making sure that peer review aligns with the needs of the community and actually improves the quality of the work, rather than having "malicious peer review" to get some citations or pet peeves in.
Given the power that the ML field holds and the interesting experiments with open review, I would wish for the field to engage more with the scientific system at large and perhaps try to drive reforms and improve it, rather than completely abandoning it and treating a PDF hosting service as a journal (ofc, preprints would still be desirable and are important, but they can not carry the entire field alone).
Simply anticipating basic push backs from reviewers makes sure that you do a somewhat thorough job. Not 100% thorough and the reviews are sometimes frivolous and lazy and stupid. But just knowing that what you put out there has to pass the admittedly noisily gatekept gate of peer review overall improves papers in my estimation. There is also a negative side because people try to hide limitations and honest assessments and cherry pick and curate their tables more in anticipation of knee jerk reviewers but overall I think without any peer review, author culture would become much more lax and bombastic and generally trend toward engagement bait and social media attention optimized stuff.
The current balance where people wrote a paper with reviers in mind, upload it to Arxiv before the review concludes and keep it on Arxiv even if rejected is a nice balance. People get to form their own opinion on it but there is also enough self-imposed quality control on it just due to wanting it to pass peer review, that even if it doesn't pass peer review, it is still better than if people write it in a way that doesn't care or anticipate peer review. And this works because people are somewhat incentivized to get peer reviewed official publications too. But being rejected is not the end of the world either because people can already read it and build on it based on Arxiv.
The problem is that "optimizing for peer-review" is not the same thing as optimizing for quality. E.g., I like to add a few tongue-in-cheeks to entertain the reader. But then I have to worry endlessly about anal-retentive reviewers who refuse to see the big picture.
Currently a kind of rule of thumb is that a PhD student can graduate after approximately 3 papers published in a good peer reviewed venue.
If peer review were to go away, this whole academic system would get into a crisis. It's dysfunctional and has many problems but it's kinda load bearing for the system to chug along.
You'll still need to convince the purseholders to pay you, and they'll want some objective metric to measure your output, and whatever metric they pick will be gamed.
But this is already how the purse holders operate. A big group of experts get together and vote on which grant proposals within a given category to fund.
I think it comes down to how the system is structured and how many players there are. The more difficult it is for a small cult to capture control of the funding (or access to instrumentation or awarding of degrees or whatever) for a given area the less likely you are to end up with a monoculture.
Assuming the majority of the funding continues to come from governments then you have a centralized point of leverage that can shape the system. So it should be possible to impose constraints that result in a system that actively prevents monocultures from developing.
The arXiv vs journal debate seems a lot like 'should the work get done, or should the work get certified' that you see all over 'institutions', and if the certification does not actually catch frauds or errors, it's not making the foundations stronger, which is usually the only justification for the latter side.
Responding largely to the linked article, you can't just ignore the massive increase in funding and associated output that occurred. Scaling almost any system up will be expected to result in creative new failure modes. It's easy to observe that a system isn't great and suppose that removing it would improve things but this very often isn't the case. Democracy is one such example.
There's also the publishing ecosystem that developed around the increased funding. It isn't clear to me why any blame (if it's even valid, see preceding paragraph) should be laid at the feet of the practice of peer reviewing publications rather than such an obviously dysfunctional institution.
Even if we accept the way in which publications have been undergoing peer review to somehow be the root of all evil (as opposed to the for profit publication of taxpayer funded work) - there's more than one way to go about it! A glaringly obvious problem, mentioned in the linked article yet not meaningfully addressed that I saw, is that peer reviewers aren't paid. If this was a compensated task presumably it would be performed much more rigorously. Building inspectors aren't volunteers and they seem to do a good enough job.
What's the value of academic publishing over the arxiv model of freely publishing, free access, and a global, vigorous discussion across a wide range of platforms, with experts, researchers, amateurs, institutions, and the peanut gallery all having the opportunity to participate?
What possible value does a journal like Nature, for example, bring to the table by claiming a paper for themselves and charging people for it, given the alternative?
I don't see any value there. Maintaining an exclusive clique by using artificial scarcity while coasting on the dregs of reputation remaining to a once prestigious institution is what a lot of these journals are doing.
The world has changed. There's no need for that sort of pay to play gatekeeping, and in fact, the model does tremendous damage to academic and intellectual integrity. It allows people to get away with fraud and it makes the institutions motivated to hide and cover it up so as to not damage their own reputations by admitting anything slipped by them.
If you contrast the damage done by journals, with regards to suppressed research, gatekept access, money taken from researchers and readers alike, against the value they might plausibly provide, the answer is clear.
They're not needed anymore. The AI era, since 2017, has thoroughly demonstrated that journals are materially incapable of keeping up, that they're unable to meaningfully contribute to the field, and that their curation or other involvement has no effective practical value. The same is true for other fields, but everyone involved wants to keep their piece of the grift going as long as possible.
We don't need them, anymore. I suspect we never did.
The value is the ability to do science as a career without being independently wealthy.
Politicians, administrators, donors, and taxpayers don't want scientists deciding on their own how to spend the money. They want control over what gets funded. They want funding decisions with justifications they can understand. But they don't understand the science itself, so they need "objective" metrics to support the decisions. And because those metrics matter, people will inevitably game them.
This is a serious risk for the open source ecosystem and particularly the scientific ecosystem that over the last years has adopted many of these technologies. Having their future depend on a cap-ex heavy company that is currently (based on reporting) spending approx. 2.5 dollars to make a dollar of revenue and must have hypergrowth in the next years or perish is less than ideal. This should discourage anybody doing serious work to adopt more of the upcoming Astral technologies like ty and pyx. Hopefully, ruff and uv are large enough to be forked should (when) the time comes.
On the flip side, I'm not sure I ever saw a revenue plan or exit strategy for Astral other than acquihire. And most plausible bidders are unfortunate in one way or another.
Astral was building a private package hosting system for enterprise customers. That was their stated approach to becoming profitable, while continuing to fund their open source work.
It's also a crowded and super mature space space between JFrog (Artifactory) and Sonatype (Nexus). They already support private PyPI repositories and are super locked in at pretty much every enterprise-level company out there.
A commodity yes, but could be wrapped in to work very nicely with the latest and greatest in python tooling. Remember, the only 2 ways to make money are by bundling and unbundling. This seems like a pretty easy bundling story.
Yeah you'd think so but somehow JFrog (makers of Artifactory) made half a billion dollars last year. I don't really understand that. Conda also makes an implausible amount of money.
Most of the companies that spend $$$$ with them can't use public registries for production/production-adjacent workloads due to regulations and, secondarily a desire to mitigate supply chain risk.
Artifactory is a drop-in replacement for every kind of repository they'll need to work with, and it has a nice UI. They also support "pass-through" repositories that mirror the public repositories with the customization options these customers like to have. It also has image/artifact scanning, which cybersecurity teams love to use in their remediation reporting.
It's also relatively easy to spin up and scale. I don't work there, but I had to use Artifactory for a demo I built, and getting it up and running took very little time, even without AI assistance.
Yeah I mean I understand the demand. My previous company used Artifactory. I just don't understand why nobody has made a free option. It's so simple it seems like it would be a no brainer open source project.
Like, nobody really pays for web servers - there are too many good free options. They're far more complex than Artifactory.
I guess it's just that it's a product that only really appeals to private companies?
Both Artifactory and Sonatype have somewhat restricted open-source options, which is part of their "get a foot in the door" product-driven sales strategy.
There are no competing open-source projects because such projects would need to provide more value than Artifactory/Sonatype OSS, which are both already huge projects, just to be considered.
JFrog has a free version. It's called the JFrog Container Registry. Lots of features are missing and you can't use the Artifactory API that it ships with, but it's there.
There are also several free registries out there: Quay, Harbor, and Docker's own distribution. They all have paid versions, of course.
From my understanding there are a lot of companies that need their own package repositories, for a variety of reasons. I listened to a couple podcasts where Charlie Marsh outlined their plans for pyx, and why they felt their entry into that market would be profitable. My guess is that OpenAI just dangled way more money in their faces than what they were likely to get from pyx.
Having a private package index gives you a central place where all employees can install from, without having to screen what each person is installing. Also, if I remember right, there are some large AI and ML focused packages that benefit from an index that's tuned to your specific hardware and workflows.
Private artifact repositories also help to mitigate supply chain risk since you can host all of your screened packages and don't have to worry about something getting removed from mvn-central, PyPI, NPM, etc.
Plus the obvious need for a place to host proprietary internal libraries.
> a lot of companies that need their own package repositories
Every company needs its own package repository. You need to be able to control what is running on your environment. Supply-chain risk is very, very real and affects anybody selling software for a living.
This is besides the point that in the real world, not every risk is addressed, at least in part because available resources are diverted to address larger risks.
How many people use that simple pip repo daily? If the number is not in the high hundreds, or a few thousands; maybe nothing. But once you get up there, any kind of better coordination layer is useful enough to pay money to a third party for, unless maintaining a layer over pip is your core competency.
i mean ofc but like you can self-host pypi and the "Docker Hub" model isn't like VC-expected level returns especially as ECR and GHCR and the other repos exist
They could have joined projects like the Linux Foundation which try to not depend on any single donor, even though complete independence from big tech is not possible. I don't know the motivation behind Astral's approach, but this acquisition does leave a weird taste behind about how serious they were about truly open source software. Time will tell, I guess. (Edit: typo)
> I don't know the motivation behind Astral's approach, but this acquisition does leave a weird taste behind about how serious they were about truly open source software.
My hope would be that this eventually pushes pip to adopt a similar feature-set and performance improvements. It's always a better story when the built-in tool is adequate instead of having to pick something. And yes UV is rust but it's pretty clear that Python could provide something within 2-5x the speed.
There seems to be a pervasive believe that the Python tooling and interpreter suck and are slow because the maintainers don’t care, or aren’t capable.
The actual problem is that there isn’t enough money to develop all of these systems properly.
Google says that Astral had 15 team members. Or course, it’s so hard to make these projections. But it wouldn’t shock me if uv and ruff are each individually multi-million dollar pieces of software.
If you’d like to invest a million dollars to improve pip, or work for free for 3 years to do it yourself, I’m not sure if anyone would object.
`ensurepip` does not "download pip for you". It bootstraps pip from a wheel included in a standard library sub-folder, (running pip's own code from within that wheel, using Python's built-in `zipimport` functionality).
That bootstrapping process just installs the wheel's contents, no Internet connection required. (Pip does, of course, download pip for you when you run its self-upgrade — since the standard library wheel will usually be out of date).
as someone who works in the python tooling space I think you underestimate the number of people who would be willing to do this. i would personally help maintain a community fork of ruff if it got to the point where one was needed, though I draw the line at moving to nebraska first.
It is proven by the amount of projects that eventually falled by the wayside after the first wave of volunteers run out of steam to keep it going post-fork.
that is a fair point, but I also believe that that happens when the project gets superseded by something better. I do not think ruff or uv will die because people went back to earlier solutions, if openai does kill them and the community fork runs out of steam it will be because someone made an even better tool, possibly incorporating the lessons learnt from astral's efforts.
This might be true for uv and ruff, and hopefully that will happen. But pyx is a platform with associated hosting and if successful would lock people into the Astral ecosystem, even if the code itself was open source.
> I never adopted them, keep using mostly Python written stuff.
Maybe you use non-transitive pure Python dependencies, but it's likely that your tools and dependencies still rely on stuff in Rust or C (e.g.: py-cryptography and Python itself respectively).
I use mostly the batteries, given that the only purpose I have for Python, since version 1.6, is UNIX scripting tasks, beyond shell.
As mentioned multiple times, since my experience with Tcl and continuously rewriting stuff in C, I tend to avoid languages that don't come with JIT, or AOT, in the reference tooling.
I tend to work with Java, .NET, node, C++, for application code.
Naturally AI now changes that, still I tend to focus on approaches that are more classical Python with pip, venv, stuff written in C or C++ that is around for years.
There are ways to independently fund open source projects, though. I have previously contributed to the Python Software Foundation and to individual open source maintainers through GitHub donations (which are not dependent on GitHub, as there are many alternatives). Projects like the Linux Foundation exist, too. And government funding, especially for scientific endeavors or where software is used to fulfill critical state tasks, is an option, too. I refuse to subject to the hypercommercialization of software and still believe in the principles behind open source.
> This is a serious risk for the open source ecosystem and particularly the scientific ecosystem that over the last years has adopted many of these technologies.
At worst, it's just Anaconda II AI Boogaloo. The ecosystems will evolve and overcome, or will die and different ecosystems rise to meet the need going forward.
I anticipate OpenAI will get bored and ignore Astral's tools. Software entropy will do its thing and we will remember an actively developed uv as the good old days until something similar to cargo gets adopted as part of Python's standard distribution.
How much money do they make from donations? I don't know but "In practice we frequently payed for travel and hardware."
Translation: nothing at all.
If such a fundamental project that is a revenue driver for so many companies, including midas-level rich companies like Google, can't even pay decent salaries for core devs from donations, then open source model doesn't work in terms of funding the work even at the smallest possible levels of "pay a reasonable market rate for devs".
You either get people who just work for free or businesses built around free work by providing something in addition to free software (which is hard to pull off, as we've seen with Bun and Astral and Deno and Node).
There's a wide gap between the arguments "the open source model doesn't work" and "the open source model failed to produce anything as good as uv after a couple decades of python tooling churn". The latter is why people are understandably unsure of where things go from here.
Seems like you're responding to the wrong person. The person I replied to said the open source model doesn't work. Nobody said the thing in your second quote.
I get the point you're making, but the way you introduced it isn't conducive to productive conversation.
I don't agree. If anything, I'd argue that the comments above were a lot less conducive to productive conversation than mine ("Would single maintainers of critical open source projects be a better situation?", "Are you not aware of foundations?", "But sure, the entire open source model doesn't work, lol").
The entire context of this subthread is whether or not the model that Astral was using was reasonable or not compared to an open source approach. From your initial comment, you've been touting alternatives, and the comment I responded to was giving specific examples of where you think the model worked. I don't think you've provided much evidence that there was a good alternative here, and when you're taking an opinionated stance, a productive conversation will sometimes involve people pointing out flaws they perceive in your arguments.
This did happen in summer of 2022, but only in public buildings, private households were not affected, so the OP's point seems a bit overly dramatic. Given that AC usage is highest when solar production is also highest, this seems highly unlikely given the solar build-out of the last 4 years.
A 2 person startup cannot provide a dedicated developer for your account, a personal contact for each of their thousand customers, is at high risk of being acquired/changing their business model/founders abandoning it, etc.
For enterprise, long-term stability and personal contact matters more than price. A typical SaaS contract is 0.x% of yearly revenue of big corps and nobody wants to be the one person risking the business for such miniscule savings. Another often overlooked part: Employees are the biggest cost center, much larger than any contract. So retraining a single team of 10 employees can often be more expensive and more disruptive to the business than just sticking with a legacy provider and established processes.
I'm not sure this matters. Enterprise is always slow to move anyways, and frankly, not usually worth the trouble for early startups.
What happens instead is that the new cheaper competitor proves themselves in the 1-10 seat company range for a few years. Then 5 to 10 years later, when the enterprise is evaluating renewals again, they go "Why are you so much more expensive? Look "X-two-guys" over there only charge 5% as much as you for the same product!" to the current SaaS they buy from.
Will they all move? No. But enough will, eventually.
I agree. The notary process is a bit annoying, but it only costs 500-1,000 euros. Yes, that's not ideal, but if you're building a proper business, that shouldn't be an issue. You can typically get an appointment within a week, no matter where you are or where your company is registered. However, once the notary sends your documents off, it can take days or weeks for the registry courts to handle them. You have to register with ten other places yourselves, and there's no guidance. There are different forms and requirements, and the yearly costs just for a basic tax declaration are in the thousands. They can be 5-10% of early startup expenses for no good reason.
There are also some shady setups, such as a private company handling the company registry for the state. You have to pay this company each year to publish your books. Accessing that data still costs money, except for the largest companies. It doesn't help with transparency, but it is a public-private rent-seeking nightmare that (possibly) arose due to conflicts of interest among certain politicians (the company's CEO has a higher-level position in political party) and lobbying.
