The Mistral team is just as sketchy as the other AI folks, they’re former Google and Meta employees. I want to believe they won’t be like Altman or Amodei. But they are following the same playbook: in debt to investors, making promises (saying their data centre will find cures to diabetes and cancer), pirating books, and lobbying the gov’t to pay a one-time penalty to avoid future litigation in plagiarism cases…
Still, fingers crossed!
They are under the EU AI act, so that’s something.
Too bad the EU is demolishing the AI Act to appease Trump.
And anyway all other vendors would be under the EU AI Act in the EU market.
By the end of Trump, it’s doubtful anybody will be doing stuff to appease him.
By the end of Trump we should be happy if there is anybody left anywhere with significant power to even think to have a leverage against anyone else.
Even OPEC is coming apart.
I use mistral, and have been happily paying for it the past 3 months. It’s more than good enough for my needs and the only AI company I trust with my data.
Even used their coding assistant to vibe code some simple apps for my personal usage recently and was pleasantly suprised.
If you would rather not trust any AI company with your data, consider heading to !localllama@sh.itjust.works where self-hosted LLM are discussed!
Mistral open sources a lot of their models.
Le Chat and Mistral Vibe are very good products. I use Vibe on a daily basis. Works very well with my style (relatively small iterative steps, very exact prompts) and it’s often faster than the competition.
I can only agree. I choosed Mistral because EU, I stayed because pretty damn good.
My style too.
A comment on HN pointed to BullshitBench. Seems like small steps is an especially good idea.
Mistral improved by a lot.
lol at people sharding for AI flavors
stupid ‘vibes’ all the way down
what, are you not excited for the epic new trustworthy european startup Le Nexus du Tourment?
Is there a use case where Mistral still beat Qwen or Gemma? If you’re using Mistral, which model and what do you use it for?
qwen suck in Danish. Much better with Mistral.
I’m using it, all I know is it’s called “Le Chat”, is that what’s meant by model? - as you can tell I’m not very technical.
I use it for generic simple queries. No idea how it compares with the others as I’ve not used them. I’m using it, as the title suggests "they’re not American*
I see. Mistral was the favorite in self-hosted LLM circles back in 2023-2024 but general opinion is that they have since been far surpassed by Chinese and American models, hence my question.
Good to know they’ve found a market with their online offering.
I was a paying customer of Mistral for about a year, it’s quiet good… but nevertheless I moved to Google this month. Gemini is - at least in my experience - simply oh so much better…
Depends on what you are looking for. Mistral’s strong suite is that it can use most, if not all European languages. Also Mistral’s basic architecture is much more (cost & energy) effective due to the use of their use of seperate expert nodes iirc.
