China is absolutely winning innovation in the 21st century. I'm so impressed. For an example from just this morning, there was an article that they're developing thorium reactor-powered cargo ships. I'm blown away.
> The tech is from America actually, decades ago. (Thorium).
I guess it depends on how you see it, but regardless, the people putting it to use today doesn't seem to be in the US.
FWIW:
> Thorium was discovered in 1828 by the Swedish chemist Jöns Jacob Berzelius during his analysis of a new mineral [...] In 1824, after more deposits of the same mineral in Vest-Agder, Norway, were discovered [...] While thorium was discovered in 1828 its first application dates only from 1885, when Austrian chemist Carl Auer von Welsbach invented the gas mantle [...] Thorium was first observed to be radioactive in 1898, by the German chemist Gerhard Carl Schmidt
For being an American discovery, it sure has a lot of European people involved in it :) (I've said it elsewhere but worth repeating; trying to track down where a technology/invention actually comes from is a fools errand, and there is always something earlier that led to today, so doesn't serve much purpose except nationalism it seems to me).
Jm2c, but I really dislike those winners/losers narratives. They lack any nuance, are juvenile, and ultimately do not contribute much but noise like endless of pointless "who's better Jordan or Lebron?" debates.
I don't know if how close Europe is, but I'm sufficiently whelmed by Mistral that I don't need to look elsewhere yet. It's kind-of like having a Toyota Corolla while everybody else is driving around in smart cars but it gets it done. On top of it, there's a loyal community that (maybe because I'm not looking) I don't see with other products. It probably depends on your uses, but if I spent all my time chasing the latest chat models (like Kimi K2 for instance) I wouldn't actually get anything done.
> I don't know if how close Europe is, but I'm sufficiently whelmed by Mistral that I don't need to look elsewhere yet. It's kind-of like having a Toyota Corolla while everybody else is driving around in smart cars but it gets it done.
My problem was that it really doesn't, none of the models out there are that great at agentic coding when you care about maintainability. Sonnet 4.5 sometimes struggles and is only okay with some steering, same for Gemini Pro 2.5, GPT-5 recently seems like it's closer to "just working" with high reasoning, but still is expensive and slow. Cerebras recently started offering GLM-4.6 and it's roughly on par with Sonnet 4 so not great, but 24M tokens per day for 50 USD seems like good value even with 128k context limitation.
I don't think there is a single model that is good enough and dependable enough in my experience out there yet, I'll probably keep jumping around for the next 5-10 years (assuming the models keep improving until we hit diminishing returns so hard that it all evens out, hopefully after they've reached a satisfying baseline usefulness).
Don't get me wrong, all of those models can already provide value, it's just that they're pretty finnicky a lot of the time, some of it inherent due to how LLMs work, but some of it also because they should just be trained better and more. And the tools they're given should be better. And the context should be managed better. And I shouldn't see something as simple as diffs fail to apply repeatedly just because I'm asking for 100% accuracy in the search/replace to avoid them messing up the brackets or whatever else.
I use Mistral's models, I've built an entire internal-knowledge-pipeline of sort using Mistral's products (which involved anything from OCR, to summarization, to linking stuff across different services like Jira or Teams, etc) and I've been very happy with it.
We did consider alternatives and truth to be told none was as cost-effective, fast and satisfying (and also our company does not trust US AI companies to not do stuff with our data).
My god the cost right? It's so much less than any of the competition that just feeding off of an api key (for coding, yeah) works great.
But as you say the rest of it is good too. I use it for research and to me it does a great job, all for a fraction of the price and the carbon of the U.S. players.
You have to try the latest Corolla then. Really smart. Lane and collision assistance, ... Unlike my old Corolla which is total dumb. It even doesn't turn the light off when I leave the car
energy generation multiples of what the US is producing. What does AI need ? Energy.
second - the open source nature of the models - means as you said a high baseline to start with - faster iteration.