In a world of grok inserting south african genocide talking points into random unrelated questions…
A french AI that randomly throws in croissant recipies & serving suggestions could be a funny rebuke.
I suspect it’ll do so while also claiming that Haiti was always wrong and much better when it was their colony
it’s more than Haiti
I use it all the time for my llm demands (trivial explanations and other smaller prompts)
works like a charm, dont really miss other llm’s. sometimes (1 in 100+ prompts) I catch myself comparing to other llm’s for “better” answers and then always I realize this is not a good thing to ask llm’s anyways (reliability, research confirmation needed, etc)
can recommend!
I don’t know how other models maybe catched up as well, but I was surprised how well it handled math related questions. If it was too complicated it simply wrote a script which it executed itself and gave me the answer in text
Agreed, I also use their code suggestion model instead of Github Copilot since a while and it works nicely
I also have a good experience with mistral
I use it. The iOS app is bad (on my phone at least) where it often claims network error and then forget the prompt I just wrote. It’s so annoying!
Can access this via Duck AI, and compare to other models on the fly.
Mistral’s model available on Duck AI is Mistral Small 3 (there’s 3.1 already but not on Duck AI - it’s open source though), they have other models too you can check out on a page their differences
