Hacker News Reader

Comment by manlymuppet | Hacker News Reader

Man, a lot of this discussion sounds like people cheering for the last kid crossing the finish line.

Surely we want competition and Europe involved in that, but at this point I have grown used to either American labs smashing the frontier remarkably fast, or Chinese labs getting way, way closer than you would expect them to.

Mistral’s progress, regrettably, feels much slower. This model doesn’t knock anybody’s socks off. The model is (and I hate to be this harsh) mediocre, and this mediocrity has also arrived months late.

This is a pretty grim prognosis for European AI.

View on HN

> Man, a lot of this discussion sounds like people cheering for the last kid crossing the finish line.

Sometimes it's ok to cheer for the last kid crossing the finish line because they're actually running a totally different race, and winning might look completely different.

When I look at what Mistral does vs other organizations I'm impressed:

https://isaiprofitable.com/

They aren't profitable yet, but they're a lot closer than most and they're doing a hell of a lot with very little.

Pointless racing story:

I was in high school track with a really tough guy who was just not a runner. We went to a pretty messed up high school and if you screwed around in track practice sometimes the coach would make you run a crap race at the next meet, like steeplechase or hurdles. Well this guy and a few others screwed up and coach made them all run hurdles at a meet.

He hooked every single one and fell on his face. Every time he got back up and kept on running. By the time he hit the finish line his knees were bleeding halfway down to his ankles. We cheered like hell and he was smiling ear to ear.

Coach quit punishing us with races after that.

I read that site quite differently from you. You seem to be analyzing absolute differences but ROI is really about ratio of spending to revenue.

It looks like Mistral is middle of the pack, behind Anthropic and ahead of OpenAI on that front. All of those labs are way "ahead" of the cloud providers, but those providers are building infrastructure, not just training models, so it's not apples to apples.

Yeah that's a good point. The numbers are smaller and I was looking at it absolutely.

Speaking from an absolute perspective I do think they are doing more with their money than either Anthropic or OpenAI.

Mistral is more profitable as a general percentage sure, but take their overall revenue: about $400 million annually.

Good, until you compare that to OpenAI’s behemoth $70 billion.

Sure OpenAI has much higher operating expenses, but which position would you rather be in? If OpenAI cut off all training and R&D tomorrow, would their 1 billion users simply vanish? Could Mistral ever realistically acquire 1 billion users?

Point being that profitable isn’t telling the whole story. And I don’t know if lack of resources is a great excuse, when Chinese labs are doing more with less, and it’s ultimately up to every competitor to win as best they can, regardless of their circumstances.

Great story you told, though. My old track coach was also an asshole, but of the more lovable variety.

>when Chinese labs are doing more with less

I need a source for this. AFAIK, the government heavily subsidize those labs.

I think right now it should definitely be a race to profitability, because revenue will not save you from unsustainable finances, no matter how large the values are.

I think OpenAI is currently far enough ahead that they could just cut training and R&D for a while (maybe even 1-2 years) and rake in some money before they'd need to start again, but if a smaller company finds a way to keep doing R&D while also being profitable, it doesn't matter if they're a little behind in quality, they will win in the long term.

This is a great story. I don't think we should expect someone to win in the first races they participate. There are football teams that never win the championship and they always play and strive to be better. Mistral it's definitely improving with each release and props to them for releasing open models

Proportionally to income, Anthropic revenue is two thirds of its expenses, Mistral is a half, and OpenAI a little bit than a half. They aren't that different.

> This is a pretty grim prognosis for European AI.

I think it's an incomplete read. What's the point in competing for a sizeable percentage of your funding when the finish line is incrementally being moved each month? Better spend it on leapfrogs which they seem to have done.

Meanwhile Mistral have a natural ace in their pocket with respect to regulation in the form of CADA and the Cloud Sovereignty Framework. I can't think of another company that would qualify as SOV-3 under that regime

So we’re supposed to cheer on companies that make worse products and only exist due to regulatory capture now?

Political polarization is turning the world insane.

As opposed to what? Companies that steal the collective intellect and intellectual property of the entire world? Yeah, what a great alternative. A world where all of your hard work is blatantly stolen.

People were extremely dismissive of chinese models until recently. They went from 1 year behind frontier to 6 months behind frontier to 3 months behind frontier extremely fast.

To be clear I'm not trying to dismiss European AI. I am a proponent of it.

But Chinese models have very much earned their place. The same cannot be said of Europe, so far.

Most comments you've made here sound like you're dismissing European AI though

I think one can be realistic about where things stand now, while also hoping for more in the future.

Important to point out that these 'X months behind frontier' really refer to the public frontier, and not the actual frontier, which private companies are free to protect indefinitely. Perhaps open models are in actuality 18 months behind the actual frontier - how would any of us know?

Because we have access to these models and use them maybe? Because of benchmarks? It does not feel like there is much moat, look you can even use small models these days and do productive work. It is rather that private ones starting to get into diminishing return area which without moat is quite risky.

Serious people haven't been very dimissive of Chinese models since at least DeepSeek-R1 in January 2025. Sadly, we in Europe are much behind.

People are cheering for a kid that is gaining ground in an ongoing race.

Hey I’m cheering too, but this kid is less like a kid and more like a multiple gold medal winning Olympian, and our once great track star is now on his last legs.

Europe has been at the heart of technological innovation for a long time. In recent years, this feels much less so.

This is a pretty mean spirited comment that fails to recognise the realities on the ground. The US is a technological juggernaut which Europe just can't compete with, because it's not a level playing field. The US has multiple moats in the form of financing, computing power, tech talent and one language.

This is very much a welcome development and we should celebrate it.

I don't think it will matter in a year or so. We are clearly topping out on useful intelligence for an increasing amount of tasks, as demonstrated by more and more models reaching the "useful" barrier.

This barrier is not going to start moving dramatically. It will simply be mostly satisfied for most work we do. Mistral is going to get there, soonish, long before the economy takes an entirely different shape (in so far that even happens).

There will be super human intelligence tasks, tasks truly constrained by intelligence for quite a while. Those will be few and far between, relatively speaking. Mistral will have plenty of opportunity to capture the other stuff, with a fraction of the resources required that it took the frontier labs to get there first.

It’s not that I necessarily think you’re wrong—-it actually makes perfect sense to me that eventually utility from frontier models will have diminishing returns—-but this has been a talking point for a long time.

The prediction that eventually more intelligence will stop being significantly more useful has been wrong many times before at this point. We should be prepared for that not to be the case.

I think it's reasonable to expect much diminishing returns for work that we have a good sense of. Compare it to a new laptop, phone or machine in a factory: There probably are better versions than the one you are using, but how much better are they and will it help you do better work? Is it even enough to be worth switching?

(I understand that the considerations are a bit different when ostensibly the switch is as simple as changing an API call, but that's not really true, when you have the AI integrated into a larger system and need to run evals and adjust the framework around it at some places at the very least)

Then, of course, there will be entirely new work that benefits from every bit of additional intelligence, but the stuff that is useful to 99% of the economy right now is probably capped on some relatively finite amount of that (with adaption/integration going to be the bulk of work)

Like I said, I don't disagree with the reason of your argument. Rather, I doubt because of a certain epistemological humility.

We can sit here and predict things all we want, but these sort of predictions don't have a great track record. The way things actually play out with AI advancement is often different from the way our perspectives reason they will play out.

It's better then to rely on previous experience and evidence, rather than using reason alone.

Curious, why do you say the model is mediocre? I haven't tried it, so I can't pass any judgement... I've learned to distrust benchmark rankings. Are benchmarks and Artificial Analysis the yardstick you're using?

> Curious, why do you say the model is mediocre? I haven't tried it, so I can't pass any judgement.

It's nowhere mediocre.

It's toes-to-toes with GLM-5.3 which is one of the best Open Weight model available (With Kimi K3) for general reasoning.

I just runned it on code reviews right now and it was able to catch some thread safety issue than DeepSeek-4.1 didn't. And DeepSeek-4.1 is by no means a bad model.

It's better than I expected for Mistral.

That said, it hardly seems like the innovative underdog some comments here make it out to be.

GLM-5.3 and Kimi K3 released 2 and 3 months ago respectively.

For all I know anyone could train a model of similar performance by just applying already published research to a new large training run.

> That said, it hardly seems like the innovative underdog some comments here make it out to be.

It's definitively not a revolution, but it is a major jump forward compared to Mistral Medium 3.5.

Worth to notice that they explicitly said they still refining the model, so we might expect something a little bit better at the end of the month

For something trained on a limited budget and 4000 GPUs, it is quite respectable.

I'm just wondering if it's really the result of impressive innovation, in line with comments saying that Mistral might in the future beat a bankrupt OpenAI due to being more profitable or efficient.

Or if it's more a situation where I could have Opus 5.5 go through recent papers from the Chinese labs to apply these to a new large training run, resulting in a model that performs similarly to Mistral Large 4.

I honestly don't know.

I would do this calculation as some measure of innovation / resources, then probably we can put this work in context a little bit better.

And if we could do innovation / red tape, I'm sure it'd auto win :)

Any lab can make the lack of resources argument. Regardless of that though, it’s every lab’s own responsibility to do well, and to do well despite circumstances.

Is that fair? Of course not! But who really expected for everything to be perfectly fair all the time anyways?

We can do more nuanced comparisons that take resources into account. That will help us see how efficient each lab is, but it doesn’t help the labs overall. Mistral shouldn’t be held to a lower standard simply because they got outcompeted.

Sure sure but if I want to answer the question "is model quality a function of resources or there's something irreplaceable in the US and Chinese engineers that can't be replicated in Europe?" it's a good metric. And if the European Union or an European VC is going to decide whether to invest in AI in Europe or not, they'd probably like to know.

Fair point.

Yes, it arrived months late. The waiting was too long I started losing faith. But when I saw Mistral’s series funding news, I knew they are on the right track. 3 billion EUR is just too significant

Doesn't matter, it's excellent, and it's European with European inference, which solves the pains of all my clients trying to build data lakes and processes on top of it.

Nobody in the real world cares about minor benchmark differences in money losing coding agents.

And nobody in the real world is giving Altman or Musk their data.

I wish you were true - but I still meet a lot of people handling sensitive data and using free or cheap version of ChatGPT or Claude with their customer data.

I think we will see some horror stories come out with data leak in the next years.

Enterprise users are not the only users, and many, many people in the real world regularly hand their data over to Altman and Musk.

You don’t have to thrilled about this situation, but it is undoubtedly the reality.

> This is a pretty grim prognosis for European AI.

Is this better or worse than no news from European AI?

Neat. Wait 3 months for the landscape to change entirely.

LLM development is jumpy. It’s hard to extrapolate very far ahead.

I agree.

When Europe does surprise us, I will be the first to commend their progress. But until then, this is where we're at.

I thought they wouldn't release anything at all, it's not dead yet :)

Those are comments from Europe. The US is waking up now and I expect them to be much harsher.

I really want them to win as that's our last horse in the AI race, but ~200 research-oriented devs out of 1800 employees? I believe they agree it's pretty doomed and have pivoted.

What? The US has been awake for almost 7 hours now

I expect the us to be sidelining all conversations in creative ways to mention anything at all about us ai companies

First few models will always be slow improving and worse. The way to improvement is working your way through a gajillion evals [1], finding bugs, gaps, and curating training data (this part involves human design as well as raw inference compute) to fix it. This is very time intensive and can't easily be "done once and then everyone has lesser work to do" since every model is different. Well, one way to accelerate it is to simply have more compute, which mostly openai and anthropic have[2].

This is mistrals first 1T-scale model and I expect the 4th or 5th generation to be close to the best for many purposes.

[1] These evals differ from the public ones like terminal-bench, are sometimes model-specific, need real, diverse usage to actually create, and are held secretly since quality of eval is the first driver behind the next step improvement of a model.

[2] It is not close. This model was trained on less than 4k GPUs, whereas astra used north of 100k GPUs.

Mistral is not that new a player though. How can we give them this much grace when other players like xAI have done more in even less time? I don't think coddling Mistral helps them.

And to the point of scale and training cluster, so what? Not only do Chinese labs have smaller clusters with less empowered GPUs, compute is Mistral's responsibility. You can't take away from other labs just because they fulfill that responsibility better.

xAI has a lot of compute. Deepseek also has a lot, not as much though. But this is changing with their new 160k huawei ascend datacenter in inner mongolia.

The lack of compute is not really attributable in that sense to mistral. First of all it needs general investor and government willingness, which is easier in a larger economy like the US or China.

Second, you need widespread usage of your paid inference service for two reasons: one it pays off your compute cost, and two it speeds up the improvement process.

The vast majority of deepseeks paid customers are within china itself (since openai and anthropic services are not reachable from china) which gives it a market. But for someone in france, there is no reason to use a structurally slower developing model from mistral compared to using one from openai...except when data guarantees are needed, hence the landing page focus on sovereignty. As far as the dual use aspect goes, a model like this is more than enough, so the government will be happy.

My point is that pretty much anybody can make the case that getting compute is hard. Regardless of that though, it’s every lab’s own responsibility to do well, and to do well despite circumstances.

Is that fair? Of course not! But it’s not useful to remark on parity, since nobody ever expected everything to be perfectly fair anyways.

We can make all the excuses we want. All it results in is holding Mistral to a lower standard. Does that result in more fair comparisons in regard to resources to output ratio? Maybe. But in my opinion, this sentiment is detrimental to European AI overall. We shouldn’t simply accept European labs as perennial losers, or normalize lower expectations. That doesn’t help anyone.

Reminds me of Gemini 3.5 Pro

hurr durr

Reply on news.ycombinator.com